WO2022055165A1 - 포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법 - Google Patents

포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법 Download PDF

Info

Publication number
WO2022055165A1
WO2022055165A1 PCT/KR2021/011600 KR2021011600W WO2022055165A1 WO 2022055165 A1 WO2022055165 A1 WO 2022055165A1 KR 2021011600 W KR2021011600 W KR 2021011600W WO 2022055165 A1 WO2022055165 A1 WO 2022055165A1
Authority
WO
WIPO (PCT)
Prior art keywords
point cloud
cloud data
patch
information
video
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/KR2021/011600
Other languages
English (en)
French (fr)
Inventor
윤여진
박한제
오세진
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
LG Electronics Inc
Original Assignee
LG Electronics Inc
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by LG Electronics Inc filed Critical LG Electronics Inc
Priority to US18/022,900 priority Critical patent/US20230419557A1/en
Publication of WO2022055165A1 publication Critical patent/WO2022055165A1/ko
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T9/00Image coding
    • G06T9/40Tree coding, e.g. quadtree, octree
    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T17/00Three-dimensional [3D] modelling for computer graphics
    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T17/00Three-dimensional [3D] modelling for computer graphics
    • G06T17/20Finite element generation, e.g. wire-frame surface description, tesselation
    • G06T17/205Re-meshing
    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/70Determining position or orientation of objects or cameras
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/102Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
    • H04N19/124Quantisation
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/102Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
    • H04N19/129Scanning of coding units, e.g. zig-zag scan of transform coefficients or flexible macroblock ordering [FMO]
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/102Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
    • H04N19/13Adaptive entropy coding, e.g. adaptive variable length coding [AVLC] or context adaptive binary arithmetic coding [CABAC]
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/169Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
    • H04N19/18Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being a set of transform coefficients
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/50Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
    • H04N19/597Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding specially adapted for multi-view video sequence encoding
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/70Methods or arrangements for coding, decoding, compressing or decompressing digital video signals characterised by syntax aspects related to video coding, e.g. related to compression standards
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/90Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using coding techniques not provided for in groups H04N19/10-H04N19/85, e.g. fractals
    • H04N19/96Tree coding, e.g. quad-tree coding
    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/10Image acquisition modality
    • G06T2207/10028Range image; Depth image; 3D point clouds

Definitions

  • Embodiments provide Point Cloud content to provide users with various services such as VR (Virtual Reality), AR (Augmented Reality, Augmented Reality), MR (Mixed Reality), and autonomous driving service. provide a way
  • a point cloud is a set of points in 3D space. There is a problem in that it is difficult to generate point cloud data because the amount of points in 3D space is large.
  • An object of the present invention is to provide a point cloud data transmission apparatus, a transmission method, a point cloud data reception apparatus, and a reception method for efficiently transmitting and receiving a point cloud in order to solve the above-described problems.
  • An object of the present invention is to provide a point cloud data transmission apparatus, a transmission method, a point cloud data reception apparatus, and a reception method for solving latency and encoding/decoding complexity.
  • a method for transmitting point cloud data includes encoding point cloud data; and transmitting the point cloud data; may include
  • a method for receiving point cloud data may include receiving point cloud data; decoding the point cloud data; and rendering the point cloud data; may include
  • the point cloud data transmission method, the transmission device, the point cloud data reception method, and the reception device may provide a quality point cloud service.
  • the point cloud data transmission method, the transmission device, the point cloud data reception method, and the reception device may achieve various video codec schemes.
  • the point cloud data transmission method, the transmission device, the point cloud data reception method, and the reception device may provide universal point cloud content such as an autonomous driving service.
  • FIG. 1 shows an example of the structure of a transmission/reception system for providing Point Cloud content according to embodiments.
  • FIG. 2 shows an example of a point cloud data capturer according to embodiments.
  • FIG. 3 shows an example of a point cloud, a geometry, and a texture image according to embodiments.
  • FIG. 4 shows an example of V-PCC encoding processing according to the embodiments.
  • FIG. 5 shows an example of a tangent plane and a normal vector of a surface according to embodiments.
  • FIG. 6 shows an example of a bounding box of a point cloud according to embodiments.
  • FIG 7 shows an example of individual patch location determination of an occupancy map according to embodiments.
  • FIG. 8 shows an example of the relationship between normal, tangent, and bitangent axes according to embodiments.
  • FIG 9 shows an example of the configuration of the minimum mode and the maximum mode of the projection mode according to the embodiments.
  • FIG 10 shows an example of an EDD code according to embodiments.
  • FIG. 11 shows an example of recoloring using color values of adjacent points according to embodiments.
  • FIG. 13 shows an example of a possible traversal order for a block of 4*4 size according to embodiments.
  • FIG. 15 shows an example of a 2D video/image encoder according to embodiments.
  • V-PCC decoding process shows an example of a V-PCC decoding process according to embodiments.
  • FIG 17 shows an example of a 2D video/image decoder according to embodiments.
  • FIG. 18 shows an example of an operation flowchart of a transmitting apparatus according to the embodiments.
  • FIG. 19 shows an example of an operation flowchart of a receiving apparatus according to the embodiments.
  • FIG. 20 shows an example of a structure capable of interworking with a method/device for transmitting and receiving point cloud data according to embodiments.
  • FIG. 21 illustrates a voxelized surface light field sequence according to embodiments.
  • FIG. 22 shows an example in which 2D point cloud data according to embodiments are displayed separately for each point according to activity region information.
  • FIG. 23 shows a configuration of a 5-attribute video and 1-geometry video generated from an SLF data set according to embodiments.
  • FIG. 24 shows an example of a method for selecting a camera viewpoint for each patch and generating a texture video according to embodiments.
  • 25 illustrates a texture video generation method according to embodiments.
  • 26 shows a V3C bitstream structure according to embodiments.
  • 29 is attribute information according to embodiments.
  • FIG. 30 shows a video synthesis structure according to embodiments.
  • 31-32 show atlas sequence parameter sets according to embodiments.
  • 33 shows an atlas frame parameter set according to embodiments.
  • 34-35 show atlas frame tile information according to embodiments.
  • FIG. 40 shows an apparatus for transmitting V-PCC point cloud data according to embodiments.
  • 41 shows an example of an SLF data set including multiple objects according to embodiments.
  • FIG. 48 illustrates a method of transmitting point cloud data according to embodiments.
  • 49 illustrates a method for receiving point cloud data according to embodiments.
  • FIG. 1 shows an example of the structure of a transmission/reception system for providing Point Cloud content according to embodiments.
  • Point cloud content may represent data representing an object as points, and may be referred to as point cloud, point cloud data, point cloud video data, point cloud image data, and the like.
  • Point cloud data transmission device is a point cloud video acquisition unit (Point Cloud Video Acquisition, 10001), a point cloud video encoder (Point Cloud Video Encoder, 10002), file / segment encapsulation a unit 10003 and/or a transmitter (or Communication module) 10004 .
  • the transmission device may secure, process, and transmit a point cloud video (or point cloud content).
  • the transmission device includes a fixed station, a base transceiver system (BTS), a network, an Ariticial Intelligence (AI) device and/or system, a robot, an AR/VR/XR device and/or a server, and the like. can do.
  • the transmission device 10000 uses a radio access technology (eg, 5G NR (New RAT), LTE (Long Term Evolution)) to communicate with a base station and/or other wireless devices; It may include robots, vehicles, AR/VR/XR devices, mobile devices, home appliances, Internet of Things (IoT) devices, AI devices/servers, and the like.
  • a radio access technology eg, 5G NR (New RAT), LTE (Long Term Evolution)
  • 5G NR New RAT
  • LTE Long Term Evolution
  • IoT Internet of Things
  • a point cloud video acquisition unit (Point Cloud Video Acquisition, 10001) according to embodiments acquires a Point Cloud video through a process of capturing, synthesizing, or generating a Point Cloud video.
  • a Point Cloud Video Encoder 10002 encodes point cloud video data.
  • the point cloud video encoder 10002 may be referred to as a point cloud encoder, a point cloud data encoder, an encoder, or the like.
  • point cloud compression coding (encoding) according to the embodiments is not limited to the above-described embodiments.
  • the point cloud video encoder may output a bitstream including encoded point cloud video data.
  • the bitstream may include not only the encoded point cloud video data, but also signaling information related to encoding of the point cloud video data.
  • the encoder may support both a Geometry-based Point Cloud Compression (G-PCC) encoding method and/or a Video-based Point Cloud Compression (V-PCC) encoding method.
  • the encoder may encode a point cloud (referring to point cloud data or both points) and/or signaling data related to the point cloud. Specific operations of encoding according to embodiments will be described below.
  • V-PCC Video-based Point Cloud Compression
  • V-PCC Visual Volumetric Video- based Coding
  • the file/segment encapsulation module 10003 encapsulates point cloud data in a file and/or segment form.
  • a method/apparatus for transmitting point cloud data may transmit point cloud data in the form of files and/or segments.
  • a transmitter (or Communication module) 10004 transmits encoded point cloud video data in the form of a bitstream.
  • the file or segment may be transmitted to a receiving device through a network or stored in a digital storage medium (eg, USB, SD, CD, DVD, Blu-ray, HDD, SSD, etc.).
  • the transmitter according to the embodiments may communicate with a receiving device (or a receiver) through wired/wireless communication through a network such as 4G, 5G, 6G, etc.
  • the transmitter may communicate with a network system (eg, 4G, 5G, 6G, etc.) a data processing operation required according to the network system)
  • the transmission device may transmit encapsulated data according to an on demand method.
  • Point cloud data receiving device (Reception device, 10005) is a receiver (Receiver, 10006), a file / segment decapsulation unit (10007), a point cloud video decoder (Point Cloud Decoder, 10008), and / or Contains Renderer (10009).
  • the receiving device uses a radio access technology (eg, 5G NR (New RAT), LTE (Long Term Evolution)) to communicate with a base station and/or other wireless device, a device, a robot, a vehicle, AR/VR/XR devices, mobile devices, home appliances, Internet of Things (IoT) devices, AI devices/servers, etc. may be included.
  • 5G NR New RAT
  • LTE Long Term Evolution
  • a receiver 10006 receives a bitstream including point cloud video data. According to embodiments, the receiver 10006 may transmit feedback information to the point cloud data transmission device 10000 .
  • the file/segment decapsulation module 10007 decapsulates a file and/or a segment including point cloud data.
  • the decapsulation unit according to the embodiments may perform a reverse process of the encapsulation process according to the embodiments.
  • the point cloud video decoder (Point Cloud Decoder, 10007) decodes the received point cloud video data.
  • the decoder according to the embodiments may perform the reverse process of encoding according to the embodiments.
  • a renderer (Renderer, 10007) renders the decoded point cloud video data.
  • the renderer 10007 may transmit feedback information obtained from the receiving end to the point cloud video decoder 10006 .
  • Point cloud video data may transmit feedback information to a receiver.
  • the feedback information received by the point cloud transmission apparatus may be provided to the point cloud video encoder.
  • the feedback information is information for reflecting the interactivity with the user who consumes the point cloud content, and includes user information (eg, head orientation information, viewport information, etc.).
  • user information eg, head orientation information, viewport information, etc.
  • the feedback information is provided by the content transmitting side (eg, the transmission device 10000) and/or the service provider can be passed on to According to embodiments, the feedback information may be used by the receiving device 10005 as well as the transmitting device 10000 or may not be provided.
  • the head orientation information is information about the user's head position, direction, angle, movement, and the like.
  • the reception apparatus 10005 may calculate viewport information based on head orientation information.
  • the viewport information is information about the area of the point cloud video that the user is looking at.
  • a viewpoint is a point at which a user is watching a point cloud video, and may mean a central point of the viewport area. That is, the viewport is an area centered on a viewpoint, and the size and shape of the area may be determined by a Field Of View (FOV).
  • FOV Field Of View
  • the reception device 10004 may extract viewport information based on a vertical or horizontal FOV supported by the device in addition to the head orientation information.
  • the receiving device 10005 checks a user's point cloud consumption method, a point cloud video area that the user gazes at, a gaze time, and the like by performing a gaze analysis or the like.
  • the receiving device 10005 may transmit feedback information including the result of the gaze analysis to the transmitting device 10000 .
  • Feedback information may be obtained during rendering and/or display.
  • Feedback information may be secured by one or more sensors included in the receiving device 10005 .
  • the feedback information may be secured by the renderer 10009 or a separate external element (or device, component, etc.).
  • a dotted line in FIG. 1 shows a process of transferring the feedback information secured by the renderer 10009 .
  • the point cloud content providing system may process (encode/decode) the point cloud data based on the feedback information. Accordingly, the point cloud video data decoder 10008 may perform a decoding operation based on the feedback information. Also, the receiving device 10005 may transmit feedback information to the transmitting device. The transmitting device (or the point cloud video data encoder 10002) may perform an encoding operation based on the feedback information. Therefore, the point cloud content providing system does not process (encode/decode) all point cloud data, but efficiently processes necessary data (for example, point cloud data corresponding to the user's head position) based on the feedback information, and the user can provide point cloud content to
  • the transmitting apparatus 10000 may be referred to as an encoder, a transmitting device, a transmitter, etc.
  • the receiving apparatus 10004 may be referred to as a decoder, a receiving device, a receiver, or the like.
  • Point cloud data (processed in a series of acquisition/encoding/transmission/decoding/rendering) processed in the point cloud content providing system of FIG. 1 according to embodiments may be referred to as point cloud content data or point cloud video data.
  • the point cloud content data may be used as a concept including metadata or signaling information related to the point cloud data.
  • the elements of the point cloud content providing system shown in FIG. 1 may be implemented by hardware, software, a processor and/or a combination thereof.
  • the embodiments provide a user with various services such as VR (Virtual Reality), AR (Augmented Reality), MR (Mixed Reality), and autonomous driving service.
  • Point Cloud content can provide
  • a Point Cloud video may be acquired first.
  • the acquired Point Cloud video is transmitted through a series of processes, and the receiving end can process the received data back into the original Point Cloud video and render it. This allows Point Cloud video to be presented to users.
  • the embodiments provide methods necessary for effectively performing such a series of processes.
  • the whole process (point cloud data transmission method and/or point cloud data reception method) for providing the Point Cloud content service may include an acquisition process, an encoding process, a transmission process, a decoding process, a rendering process, and/or a feedback process. there is.
  • a process of providing point cloud content may be referred to as a point cloud compression process.
  • the point cloud compression process may refer to a geometry-based point cloud compression process.
  • Each element of the point cloud data transmitting apparatus and the point cloud data receiving apparatus may mean hardware, software, a processor, and/or a combination thereof.
  • a Point Cloud video may be acquired first.
  • the acquired Point Cloud video is transmitted through a series of processes, and the receiving end can process the received data back into the original Point Cloud video and render it.
  • This allows Point Cloud video to be presented to users.
  • the present invention provides a method necessary for effectively performing such a series of processes.
  • the whole process for providing the Point Cloud content service may include an acquisition process, an encoding process, a transmission process, a decoding process, a rendering process, and/or a feedback process.
  • the Point Cloud Compression system may include a transmitting device and a receiving device.
  • the transmitting device can output the bitstream by encoding the Point Cloud video, and it can be delivered to the receiving device in the form of a file or streaming (streaming segment) through a digital storage medium or network.
  • the digital storage medium may include a variety of storage media such as USB, SD, CD, DVD, Blu-ray, HDD, and SSD.
  • the transmission device may schematically include a Point Cloud video acquisition unit, a Point Cloud video encoder, a file/segment encapsulation unit, and a transmission unit.
  • the receiving device may schematically include a receiving unit, a file/segment decapsulation unit, a Point Cloud video decoder, and a renderer.
  • the encoder may be called a Point Cloud video/video/picture/frame encoding device, and the decoder may be called a Point Cloud video/video/picture/frame decoding device.
  • the transmitter may be included in the Point Cloud video encoder.
  • the receiver may be included in the Point Cloud video decoder.
  • the renderer may include a display unit, and the renderer and/or the display unit may be configured as a separate device or external component.
  • the transmitting device and the receiving device may further include separate internal or external modules/units/components for the feedback process.
  • the operation of the receiving device may follow a reverse process of the operation of the transmitting device.
  • the Point Cloud video acquisition unit may perform the process of acquiring Point Cloud video through capturing, synthesizing, or generating Point Cloud video.
  • 3D position (x, y, z)/property (color, reflectance, transparency, etc.) data for a plurality of Points are generated by the acquisition process, for example, PLY (Polygon File format or the Stanford Triangle format) file can be
  • PLY Polygon File format or the Stanford Triangle format
  • metadata related to the point cloud eg, metadata related to capture, etc.
  • metadata related to the point cloud eg, metadata related to capture, etc.
  • An apparatus for transmitting point cloud data includes an encoder for encoding point cloud data; and a transmitter for transmitting point cloud data; may include In addition, it may be transmitted in the form of a bit stream including a point cloud.
  • An apparatus for receiving point cloud data includes a receiver configured to receive point cloud data; a decoder for decoding point cloud data; and a renderer that renders the point cloud data; may include
  • a method/apparatus represents an apparatus for transmitting point cloud data and/or an apparatus for receiving point cloud data.
  • FIG. 2 shows an example of a point cloud data capturer according to embodiments.
  • Point cloud data may be acquired by a camera or the like.
  • a capture method according to embodiments may include, for example, inward-pacing and/or outward-pacing.
  • one or more cameras may photograph an object of point cloud data from the outside to the inside.
  • one or more cameras may photograph the object of the point cloud data from the inside to the outside.
  • Point cloud data or point cloud content may be a video or still image of an object/environment expressed in various types of 3D space.
  • the point cloud content may include video/audio/images for an object (object, etc.).
  • Point cloud content capture it can be composed of a combination of camera equipment that can acquire depth (a combination of an infrared pattern projector and an infrared camera) and RGB cameras that can extract color information corresponding to depth information.
  • depth information can be extracted through LiDAR using a radar system that measures the position coordinates of a reflector by emitting a laser pulse and measuring the time it takes to reflect and return. It is possible to extract the shape of a geometry composed of points in a three-dimensional space from the depth information, and extract an attribute representing the color/reflection of each point from the RGB information.
  • Point Cloud contents may consist of position (x, y, z) and color (YCbCr or RGB) or reflectance (r) information for points.
  • Point Cloud content may have an outward-facing method for capturing the external environment and an inward-facing method for capturing a central object.
  • an object e.g., a core object such as a character, player, object, actor, etc.
  • the configuration of the capture camera is based on the inward-facing method. can be used
  • the configuration of the capture camera may use an outward-facing method. Since Point Cloud content can be captured through multiple cameras, it may be necessary to calibrate the camera before capturing the content to set the global coordinate system between the cameras.
  • the Point Cloud content may be a video or still image of an object/environment displayed on various types of 3D space.
  • Point Cloud video can be synthesized based on the captured Point Cloud video.
  • capture through a real camera may not be performed. In this case, the process of simply generating related data may be substituted for the process of capturing.
  • Captured Point Cloud video may require post-processing to improve the quality of the content.
  • the point cloud extracted from the cameras sharing the spatial coordinate system can be integrated into one content through the conversion process to the global coordinate system for each point based on the position coordinates of each camera obtained through the calibration process. Through this, one wide range of Point Cloud contents can be created, or Point Cloud contents with a high density of points can be obtained.
  • a Point Cloud video encoder can encode an input Point Cloud video into one or more video streams.
  • One video may include a plurality of frames, and one frame may correspond to a still image/picture.
  • Point Cloud video may include Point Cloud video/frame/picture/video/audio/image, etc., and Point Cloud video may be used in combination with Point Cloud video/frame/picture.
  • the Point Cloud video encoder may perform a Video-based Point Cloud Compression (V-PCC) procedure.
  • the Point Cloud video encoder can perform a series of procedures such as prediction, transformation, quantization, and entropy coding for compression and coding efficiency.
  • the encoded data (encoded video/image information) may be output in the form of a bitstream.
  • the Point Cloud video encoder divides the Point Cloud video into geometry video, attribute video, occupancy map video, and auxiliary information, as described below, to encode.
  • a geometry video may include a geometry image
  • an attribute video may include an attribute image
  • an occupancy map video may include an occupancy map image.
  • the additional information may include auxiliary patch information.
  • the attribute video/image may include a texture video/image.
  • the encapsulation processing unit may encapsulate the encoded Point cloud video data and/or Point cloud video related metadata in the form of a file or the like.
  • the point cloud video-related metadata may be delivered from a metadata processing unit, etc.
  • the metadata processing unit may be included in the point cloud video encoder, or may be configured as a separate component/module.
  • the encapsulation processing unit may encapsulate the data in a file format such as ISOBMFF or process the data in the form of other DASH segments.
  • the encapsulation processing unit may include point cloud video related metadata in a file format according to an embodiment.
  • Point cloud video metadata may be included, for example, in boxes of various levels in the ISOBMFF file format, or as data in separate tracks within the file.
  • the encapsulation processing unit may encapsulate the point cloud video-related metadata itself into a file.
  • the transmission processing unit may apply processing for transmission to the encapsulated Point cloud video data according to the file format.
  • the transmission processing unit may be included in the transmission unit, or may be configured as a separate component/module.
  • the transmission processing unit can process the point cloud video video data according to any transmission protocol.
  • the processing for transmission may include processing for transmission through a broadcasting network and processing for transmission through a broadband.
  • the transmission processing unit may receive not only the point cloud video data but also the point cloud video-related metadata from the metadata processing unit, and may apply processing for transmission thereto.
  • the transmitting unit 10004 may transmit encoded video/image information or data output in the form of a bitstream to the receiving unit of the receiving device through a digital storage medium or a network in a file or streaming form.
  • the digital storage medium may include a variety of storage media such as USB, SD, CD, DVD, Blu-ray, HDD, and SSD.
  • the transmission unit may include an element for generating a media file through a predetermined file format, and may include an element for transmission through a broadcast/communication network.
  • the receiver may extract the bitstream and transmit it to the decoding device.
  • the receiver 10003 may receive the point cloud video data transmitted by the point cloud video transmission device according to the present invention. Depending on the transmitted channel, the receiver may receive point cloud video data through a broadcasting network or may receive point cloud video data through broadband. Alternatively, point cloud video data may be received through a digital storage medium.
  • the reception processing unit may perform processing according to the transmission protocol on the received point cloud video data.
  • the reception processing unit may be included in the reception unit, or may be configured as a separate component/module.
  • the reception processing unit may perform the reverse process of the above-described transmission processing unit so that the transmission side corresponds to the processing performed for transmission.
  • the reception processing unit may transmit the acquired point cloud video data to the decapsulation processing unit, and the acquired point cloud video related metadata may be transmitted to the metadata parser.
  • the point cloud video-related metadata acquired by the reception processing unit may be in the form of a signaling table.
  • the decapsulation processing unit may decapsulate the point cloud video data in the form of a file received from the reception processing unit.
  • the decapsulation processing unit may decapsulate the files according to ISOBMFF and the like to obtain a point cloud video bitstream or point cloud video related metadata (metadata bitstream).
  • the acquired point cloud video bitstream can be delivered to the point cloud video decoder, and the acquired point cloud video related metadata (metadata bitstream) can be delivered to the metadata processing unit.
  • a point cloud video bitstream may include metadata (metadata bitstream).
  • the metadata processing unit may be included in the point cloud video decoder, or may be configured as a separate component/module.
  • the point cloud video-related metadata acquired by the decapsulation processing unit may be in the form of a box or track in a file format. If necessary, the decapsulation processing unit may receive metadata required for decapsulation from the metadata processing unit.
  • the point cloud video-related metadata may be transmitted to the point cloud video decoder and used in the point cloud video decoding procedure, or may be transmitted to the renderer and used in the point cloud video rendering procedure.
  • the Point Cloud video decoder can decode the video/image by receiving the bitstream and performing an operation corresponding to the operation of the Point Cloud video encoder.
  • the Point Cloud video decoder can decode the Point Cloud video by dividing it into a geometry video, an attribute video, an occupancy map video, and auxiliary information, as will be described later.
  • a geometry video may include a geometry image
  • an attribute video may include an attribute image
  • an occupancy map video may include an occupancy map image.
  • the additional information may include auxiliary patch information.
  • the attribute video/image may include a texture video/image.
  • the 3D geometry is reconstructed using the decoded geometry image, the occupancy map, and additional patch information, and may then be subjected to a smoothing process.
  • a color point cloud image/picture may be restored by giving a color value to the smoothed 3D geometry using a texture image.
  • the renderer can render the restored geometry and color point cloud image/picture.
  • the rendered video/image may be displayed through the display unit. The user can view all or part of the rendered result through a VR/AR display or a general display.
  • the feedback process may include a process of transferring various feedback information that may be obtained in the rendering/display process to the transmitter or to the decoder of the receiver. Interactivity can be provided in Point Cloud video consumption through the feedback process.
  • head orientation information, viewport information indicating an area the user is currently viewing, and the like may be transmitted.
  • the user may interact with those implemented in the VR/AR/MR/autonomous driving environment. In this case, information related to the interaction may be transmitted to the transmitting side or the service provider side in the feedback process. there is. Depending on the embodiment, the feedback process may not be performed.
  • the head orientation information may refer to information about the user's head position, angle, movement, and the like. Based on this information, information about the area the user is currently viewing within the Point Cloud video, that is, viewport information can be calculated.
  • the viewport information may be information about the area currently being viewed by the user in the Point Cloud video. Through this, a Gaze Analysis is performed, and it is also possible to check how the user consumes the Point Cloud video, which area of the Point Cloud video how much, and so on. Gaze analysis may be performed at the receiving side and transmitted to the transmitting side through a feedback channel.
  • a device such as a VR/AR/MR display may extract a viewport area based on a user's head position/direction, a vertical or horizontal FOV supported by the device, and the like.
  • the above-described feedback information may be consumed at the receiving side as well as being transmitted to the transmitting side. That is, a decoding and rendering process of the receiving side may be performed using the above-described feedback information. For example, using head orientation information and/or viewport information, only the Point Cloud video for the region currently being viewed by the user may be preferentially decoded and rendered.
  • a viewport or a viewport area may mean an area that a user is viewing in a Point Cloud video.
  • a viewpoint is a point at which a user is watching a Point Cloud video, and may mean a central point of the viewport area. That is, the viewport is an area centered on the viewpoint, and the size and shape of the area may be determined by the Field Of View (FOV).
  • FOV Field Of View
  • This article is about Point Cloud video compression, as mentioned above.
  • the method/embodiment disclosed in this document may be applied to a point cloud compression or point cloud coding (PCC) standard of Moving Picture Experts Group (MPEG) or a next-generation video/image coding standard.
  • PCC point cloud compression or point cloud coding
  • MPEG Moving Picture Experts Group
  • a picture/frame may generally mean a unit representing one image in a specific time period.
  • a pixel or pel may mean a minimum unit constituting one picture (or image). Also, a 'sample' may be used as a term corresponding to a pixel.
  • a sample may generally represent a pixel or a value of a pixel, may represent only a pixel/pixel value of a luma component, may represent only a pixel/pixel value of a chroma component, or a depth component It may represent only the pixel/pixel value of .
  • a unit may represent a basic unit of image processing.
  • the unit may include at least one of a specific region of a picture and information related to the region.
  • a unit may be used interchangeably with terms such as a block or an area in some cases.
  • the MxN block may include samples (or sample arrays) or a set (or arrays) of transform coefficients including M columns and N rows.
  • FIG. 3 shows an example of a point cloud, a geometry, and a texture image according to embodiments.
  • the point cloud according to the embodiments may be input to the V-PCC encoding process of FIG. 4 to be described later to generate a geometry image and a texture image.
  • the point cloud may be used as the same meaning as the point cloud data.
  • the left side is a point cloud, which indicates a point cloud in which an object is located in a 3D space and can be represented by a bounding box or the like.
  • the middle represents the geometry
  • the right represents the texture image (non-padding).
  • V-PCC Video-based Point Cloud Compression
  • HEVC High Efficiency Video Coding
  • VVC Video-based Point Cloud Compression
  • occupancy map When the points constituting the point cloud are divided into patches and mapped to the 2D plane, a binary map that indicates whether data exists at the corresponding position on the 2D plane as a value of 0 or 1 (binary map) indicates An occupancy map may indicate a 2D array corresponding to an atlas, and a value of the occupancy map may indicate whether each sample position in the atlas corresponds to a 3D point.
  • An atlas is a set of 2D bounding boxes and related information located in a rectangular frame corresponding to a 3D bounding box in a 3D space in which volume metric data is rendered.
  • An atlas bitstream is a bitstream for one or more atlas frames constituting an atlas and related data.
  • An atlas frame is a 2D rectangular arrangement of atlas samples onto which patches are projected.
  • An atlas sample is a position of a rectangular frame from which patches associated with an atlas are projected.
  • An atlas frame may be divided into tiles.
  • a tile is a unit for dividing a 2D frame. That is, a tile is a unit for dividing signaling information of point cloud data called atlas.
  • Patch A set of points constituting a point cloud. Points belonging to the same patch are adjacent to each other in 3D space, indicating that they are mapped in the same direction among the six bounding box planes in the mapping process to a 2D image.
  • Geometry image An image in the form of a depth map that expresses the geometry of each point constituting the point cloud in units of patches.
  • a geometry image may consist of pixel values of one channel.
  • Geometry represents a set of coordinates associated with a point cloud frame.
  • Texture image An image that expresses color information of each point constituting a point cloud in units of patches.
  • the texture image may be composed of pixel values of multiple channels (e.g. 3 channels R, G, B). Textures are included as attributes. According to embodiments, a texture and/or an attribute may be interpreted as the same object and/or containment relationship.
  • Auxiliary patch info Represents metadata required to reconstruct a point cloud from individual patches.
  • the utility patch info may include information on the location and size of the patch in 2D/3D space.
  • V-PCC components may include an atlas, an accumulatory map, a geometry, an attribute, and the like.
  • An atlas represents a set of 2D bounding boxes. It may be a patch, for example, patches projected on a rectangular frame. In addition, it may correspond to a 3D bounding box in 3D space, and may represent a subset of point clouds.
  • Attribute represents a scalar or vector associated with each point in the point cloud, for example, color, reflectance, surface normal, time stamps, material. There may be an ID (material ID) or the like.
  • Point cloud data represents PCC data according to a video-based point cloud compression (V-PCC) method.
  • the point cloud data may include a plurality of components. For example, it may include accumulatory maps, patches, geometries and/or textures, and the like.
  • FIG. 4 shows an example of V-PCC encoding processing according to the embodiments.
  • the figure shows the V-PCC encoding process for generating and compressing an occupancy map, a geometry image, a texture image, and auxiliary patch information.
  • the V-PCC encoding process of FIG. 4 may be processed by the point cloud video encoder 10002 of FIG. 1 .
  • Each component of FIG. 4 may be implemented by software, hardware, a processor, and/or a combination thereof.
  • a patch generation (40000) or patch generator receives a point cloud frame (which may be in the form of a bitstream containing point cloud data).
  • the patch generation unit 40000 generates a patch from point cloud data.
  • patch information including information on patch generation is generated.
  • Patch packing (40001) or patch packer packs patches for point cloud data. For example, one or more patches may be packed. In addition, an accumulatory map including information on patch packing is generated.
  • a geometry image generation (40002) or geometry image generator generates a geometry image based on point cloud data, patches, and/or packed patches.
  • the geometry image refers to data including geometry related to point cloud data.
  • a texture image generation (40003) or texture image generator generates a texture image based on point cloud data, patches, and/or packed patches.
  • a texture image may be generated further based on a smoothed geometry generated by performing a smoothing (number) smoothing process on the reconstructed (reconstructed) geometry image based on patch information.
  • a smoothing (40004) or smoother may mitigate or remove errors contained in image data.
  • a smoothed geometry may be generated by gently filtering the reconstructed geometry image based on the patch information to gently filter a portion that may cause an error between data.
  • auxillary patch info compression (40005) or auxillary patch information compressor compresses additional patch information related to patch information generated in a patch generation process.
  • the compressed oscillation patch information may be transmitted to the multiplexer, and the geometry image generation 40002 may also use the oscillation patch information.
  • Image padding (40006, 40007) or image padding may pad a geometry image and a texture image, respectively. Padding data may be padded to the geometry image and the texture image.
  • a group dilation (40008) or group delimiter may append data to a textured image, similar to image padding. Additional data may be inserted into the texture image.
  • a video compression (40009, 40010, 40011) or video compressor may compress a padded geometry image, a padded texture image, and/or an accumulatory map, respectively. Compression may encode geometry information, texture information, accumulatory information, and the like.
  • the entropy compression (40012) or entropy compressor may compress (eg, encode) the accumulatory map based on an entropy scheme.
  • entropy compression and/or video compression may be respectively performed according to a case in which point cloud data is lossless and/or lossy.
  • a multiplexer 40013 multiplexes the compressed geometry image, the compressed texture image, and the compressed accumulatory map into a bitstream.
  • the patch generation process refers to a process of dividing the point cloud into patches, which are units for performing mapping, in order to map the point cloud to a 2D image.
  • the patch generation process can be divided into three steps: normal value calculation, segmentation, and patch division as follows.
  • FIG. 5 shows an example of a tangent plane and a normal vector of a surface according to embodiments.
  • the surface of FIG. 5 is used as follows in the patch generation process 40000 of the V-PCC encoding process of FIG.
  • Each point (eg, point) constituting the point cloud has its own direction, which is expressed as a three-dimensional vector called normal.
  • the tangent plane and normal vector of each point constituting the surface of the point cloud as shown in the drawing can be obtained by using the neighbors of each point obtained using a K-D tree, etc.
  • the search range in the process of finding adjacent points can be defined by the user.
  • Tangent plane A plane passing through a point on the surface and completely containing the tangent to the curve on the surface.
  • FIG. 6 shows an example of a bounding box of a point cloud according to embodiments.
  • patch generation may use a bounding box in a process of generating a patch from point cloud data.
  • the bounding box refers to a box of units for dividing point cloud data based on a hexahedron in 3D space.
  • the bounding box may be used in the process of projecting an object that is a target of point cloud data on the plane of each cube based on the cube in 3D space.
  • the bounding box may be generated and processed by the point cloud video acquisition unit 10000 and the point cloud video encoder 10002 of FIG. 1 .
  • patch generation 40000, patch packing 40001, geometry image generation 40002, and texture image generation 40003 of the V-PCC encoding process of FIG. 2 may be performed.
  • Segmentation consists of two processes: initial segmentation and refine segmentation.
  • the point cloud encoder 10002 projects a point onto one side of a bounding box. Specifically, each point constituting the point cloud is projected onto one of the faces of the six bounding box surrounding the point cloud as shown in the figure. Initial segmentation is the process of determining one of the planes of the bounding box to which each point is projected. am.
  • the normal value ( )class The plane with the largest dot product is determined as the projection plane of that plane. That is, the plane with the normal in the direction most similar to the normal of the point is determined as the projection plane of the point.
  • the determined plane may be identified as a value (cluster index) of one of 0 to 5 in the form of an index.
  • Refine segmentation is a process of improving the projection plane of each point constituting the point cloud determined in the initial segmentation process in consideration of the projection plane of adjacent points.
  • the projection plane of the current point and the projection plane of the adjacent points together with the score normal that is similar to the normal value of each point and the normal value of each plane of the bounding box considered for the projection plane determination in the initial segmentation process earlier.
  • Score smooth which indicates the degree of agreement with , can be considered at the same time.
  • Score smooth can be considered by assigning weights to the score normal, and in this case, the weight value can be defined by the user. Refine segmentation may be repeatedly performed, and the number of repetitions may also be defined by the user.
  • Patch segmentation is a process of dividing the entire point cloud into patches, which are sets of adjacent points, based on the projection plane information of each point constituting the point cloud obtained in the initial/refine segmentation process. Patch division can be composed of the following steps.
  • the size of each patch and the occupancy map, geometry image, and texture image for each patch are determined.
  • FIG 7 shows an example of individual patch location determination of an occupancy map according to embodiments.
  • the point cloud encoder 10002 may generate a patch packing and accumulatory map.
  • This process is a process of determining the positions of individual patches in a 2D image in order to map the previously divided patches to a single 2D image.
  • the occupancy map is one of the 2D images, and is a binary map that indicates whether data exists in the corresponding location with a value of 0 or 1.
  • the occupancy map consists of blocks, and the resolution can be determined according to the size of the block. For example, when the block size is 1*1, it has a pixel unit resolution.
  • the size of the block (occupancy packing block size) may be determined by the user.
  • the process of determining the location of an individual patch in the occupancy map can be configured as follows.
  • the (x, y) coordinate value of the patch occupancy map is 1 (data exists at that point in the patch), and the (u+x, v+y) coordinates of the entire occupancy map If the value is 1 (when the occupancy map is filled by the previous patch), change the (x, y) position in raster order and repeat the process of 34. If not, proceed with step 6.
  • Occupancy SizeU Indicates the width of the occupancy map, and the unit is the occupancy packing block size.
  • occupancySizeV Indicates the height of the occupancy map, and the unit is the occupancy packing block size.
  • Patch size U0 (patch.sizeU0): Indicates the width of the occupancy map, and the unit is the occupancy packing block size.
  • Patch size V0 (patch.sizeV0): Indicates the height of the occupancy map, and the unit is the occupancy packing block size.
  • a box corresponding to a patch having a patch size in a box corresponding to an Accupansa packing size block exists, and points (x, y) in the box may be located.
  • FIG. 8 shows an example of the relationship between normal, tangent, and bitangent axes according to embodiments.
  • the point cloud encoder 10002 may generate a geometry image.
  • the geometric image means image data including geometry information of the point cloud.
  • the process of generating a geometric image may use three axes (normal, tangent, and bitangent) of the patch of FIG. 8 .
  • the depth values constituting the geometry image of each patch are determined, and the entire geometry image is created based on the location of the patch determined in the patch packing process.
  • the process of determining the depth values constituting the geometry image of an individual patch can be configured as follows.
  • the parameters related to the location and size of individual patches are calculated.
  • the parameters may include the following information.
  • the tangent axis is the axis that coincides with the horizontal (u) axis of the patch image among the axes perpendicular to the normal
  • the bitangent axis is the vertical (vertical) axis of the patch image among the axes perpendicular to the normal.
  • FIG 9 shows an example of the configuration of the minimum mode and the maximum mode of the projection mode according to the embodiments.
  • the point cloud encoder 10002 may perform a patch-based projection to generate a geometry image, and modes of projection according to embodiments include a minimum mode and a maximum mode.
  • 3D spatial coordinates of the patch It can be calculated through the smallest size bounding box surrounding the patch. For example, the minimum value in the tangent direction of the patch (patch 3d shift tangent axis), the minimum value in the bitangent direction of the patch (patch 3d shift bitangent axis), the minimum value in the normal direction of the patch (patch 3d shift normal axis), etc. may be included.
  • 2D size of patch Shows the horizontal and vertical size of the patch when it is packed into a 2D image.
  • the horizontal size (patch 2d size u) is the difference between the maximum and minimum values in the tangent direction of the bounding box
  • the vertical size (patch 2d size v) is the difference between the maximum and minimum values in the bitangent direction of the bounding box.
  • the projection mode may be one of a min mode and a max mode.
  • the geometry information of the patch is expressed as a depth value.
  • the minimum depth is configured in d0 as shown in the figure, and the maximum depth existing within the surface thickness from the minimum depth can be configured as d1.
  • the point cloud when the point cloud is located in 2D as shown in the drawing, there may be a plurality of patches including a plurality of points. As shown in the drawing, the shaded points of the same style indicate that they may belong to the same patch.
  • the drawing shows the process of projecting a patch of points marked with blank spaces.
  • the depth is increased by 1, such as 0, 1, 2,..6, 7, 8, 9, based on the left, and the number for calculating the depth of the points to the right. can be indicated.
  • the same method is applied to all point clouds by user definition, or it can be applied differently for each frame or patch.
  • a projection mode capable of increasing compression efficiency or minimizing a missed point may be adaptively selected.
  • depth0 is the value obtained by subtracting the minimum value of the normal axis of each point from the minimum value of the patch normal direction (patch 3d shift normal axis) and the minimum value of the patch normal direction calculated in the process 1 (patch 3d shift normal axis). to compose the d0 image. If there is another depth value within the range within depth0 and surface thickness at the same location, set this value to depth1. If it does not exist, the value of depth0 is also assigned to depth1. Construct the d1 image with the Depth1 value.
  • a minimum value may be calculated (4 2 4 4 4 0 6 0 0 9 9 0 8 0).
  • a larger value among two or more points may be calculated, or if there is only one point, the value may be calculated (4 4 4 4 6 6 6 8 9 9 8 8 9) ).
  • some points may be lost in the process of coded and reconstructed points of the patch (eg, 8 points are lost in the figure).
  • Max mode it is a value obtained by subtracting the minimum value of the patch normal direction (patch 3d shift normal axis) calculated in step 1 from the maximum value of the normal axis of each point. Construct the d0 image with depth0. If there is another depth value within the range within depth0 and surface thickness at the same location, set this value to depth1. If it does not exist, the value of depth0 is also assigned to depth1. Construct the d1 image with the Depth1 value.
  • a maximum value may be calculated in determining the depth of the points of d0 (4 4 4 4 6 6 6 8 9 9 8 8 9). And, in determining the depth of the points of d1, a smaller value may be calculated among two or more points, or if there is only one point, the value may be calculated (4 2 4 4 5 6 0 6 9 9 0 8 0) ). Also, some points may be lost in the process of coded and reconstructed points of the patch (eg, 6 points are lost in the drawing).
  • the entire geometry image can be created by placing the geometry image of an individual patch created through the above process on the entire geometry image using the patch location information determined in the patch packing process.
  • the d1 layer of the generated entire geometry image can be encoded in several ways.
  • the first is a method of encoding the depth values of the previously generated d1 image as it is (absolute d1 method).
  • the second is a method of encoding a difference value between the depth value of the previously generated d1 image and the depth value of the d0 image (differential method).
  • Depth (EDD) codes may also be used.
  • FIG 10 shows an example of an EDD code according to embodiments.
  • the point cloud encoder 10002 and/or some/whole process of V-PCC encoding may encode geometric information of points based on the EOD code.
  • a point exists above the reference point, it becomes 1, and if the point does not exist, it becomes 0, so that a code may be expressed based on 4 bits.
  • Smoothing is an operation to remove discontinuities that may occur at the patch interface due to deterioration of image quality that occurs during the compression process, and may be performed by a point cloud encoder or a smoother.
  • This process can be said to be the reverse process of the geometry image creation described above.
  • the reverse process of encoding may be reconstruction.
  • the point is moved to the center of gravity of the adjacent points (located at the average x, y, z coordinates of the adjacent points). That is, it changes the geometry value. Otherwise, the previous geometry value is maintained.
  • FIG. 11 shows an example of recoloring using color values of adjacent points according to embodiments.
  • the point cloud encoder or texture image generator 40003 may generate a texture image based on recoloring.
  • the texture image creation process is similar to the geometry image creation process described above, and consists of creating a texture image of each patch and placing them in a determined position to create an entire texture image. However, in the process of creating the texture image of each patch, an image with color values (e.g. R, G, B) of the point constituting the point cloud corresponding to the location is created instead of the depth value for geometry creation.
  • color values e.g. R, G, B
  • the recoloring is based on the average of the attribute information of the closest original points to the point and/or the average of the attribute information of the closest original positions to the point to calculate a suitable color value of the changed position can do.
  • a texture image can also be created with two layers of t0/t1 like a geometry image created with two layers of d0/d1.
  • the point cloud encoder or oscillation patch information compressor may compress oscillation patch information (additional information about the point cloud).
  • the Oscilry patch information compressor compresses (compresses) the additional patch information generated in the patch generation, patch packing, and geometry generation processes described above.
  • Additional patch information may include the following parameters:
  • Cluster index that identifies the projection plane (normal)
  • 3D spatial position of the patch the tangent minimum of the patch (patch 3d shift tangent axis), the minimum of the patch (patch 3d shift bitangent axis), the minimum of the patch in the normal direction (patch 3d shift normal axis)
  • Mapping information of each block and patch Candidate index (When patches are placed in order based on the 2D spatial location and size information of the patch above, multiple patches can be duplicated mapped to one block. At this time, the mapped patches are It composes the candidate list, the index indicating which patch data of the list exists in the corresponding block), and the local patch index (the index indicating one of all patches existing in the frame).
  • Table X is a pseudo code showing the block and patch match process using the candidate list and local patch index.
  • the maximum number of candidate lists can be defined by the user.
  • Image padding and group dilation (40006, 40007, 40008)
  • the image fader according to embodiments may fill a space other than the patch area with meaningless additional data based on the push-pull background filling method.
  • Image padding is a process of filling spaces other than the patch area with meaningless data for the purpose of improving compression efficiency.
  • a method in which pixel values of columns or rows corresponding to the boundary surface inside the patch are copied to fill the empty space can be used.
  • a push-pull background filling method in which an empty space is filled with pixel values from a low-resolution image may be used in the process of gradually reducing the resolution of the non-padded image and increasing the resolution again.
  • Group dilation is a method of filling the empty space of the geometry and texture image composed of two layers d0/d1 and t0/t1. It is the process of filling in the average value of .
  • FIG. 13 shows an example of a possible traversal order for a block of 4*4 size according to embodiments.
  • the occupancy map compressor may compress the previously generated occupancy map. Specifically, there may be two methods: video compression for lossy compression and entropy compression for lossless compression. Video compression is described below.
  • the entropy compression process may be performed as follows.
  • the entry compressor may code (encode) the block based on the traversal order method as shown in the drawing.
  • the index is encoded by selecting the best traversal order having the minimum number of runs among possible traversal orders.
  • the drawing is a case where the third traversal order of FIG. 13 is selected, and in this case, since the number of runs can be minimized to 2, it can be selected as the best traversal order.
  • Video compression (40009, 40010, 40011)
  • the video compressor encodes a sequence such as a geometry image, a texture image, and an occupancy map image generated by the above-described process by using a 2D video codec such as HEVC or VVC.
  • FIG. 15 shows an example of a 2D video/image encoder according to embodiments.
  • the figure shows a schematic block diagram of a 2D video/image encoder 15000 in which encoding of a video/video signal is performed as an embodiment of the above-described video compression (Video compression, 40009, 40010, 40011) or a video compressor.
  • the 2D video/image encoder 15000 may be included in the above-described point cloud video encoder, or may be configured as an internal/external component.
  • Each component in Fig. 15 may correspond to software, hardware, a processor and/or a combination thereof.
  • the input image may include the above-described geometry image, texture image (attribute(s) image), occupancy map image, and the like.
  • the output bitstream (ie, point cloud video/image bitstream) of the point cloud video encoder may include output bitstreams for each input image (geometry image, texture image (attribute(s) image), occupancy map image, etc.). .
  • the inter prediction unit 15090 and the intra prediction unit 15100 may be collectively referred to as a prediction unit. That is, the prediction unit may include an inter prediction unit 15090 and an intra prediction unit 15100 .
  • the transform unit 15030 , the quantization unit 15040 , the inverse quantization unit 15050 , and the inverse transform unit 15060 may be included in a residual processing unit.
  • the residual processing unit may further include a subtraction unit 15020 .
  • the above-described image segmentation unit 15010, subtraction unit 15020, transform unit 15030, quantization unit 15040, inverse quantization unit (), ), inverse transform unit 15060, adder unit 155, filtering unit ( 15070 , the inter prediction unit 15090 , the intra prediction unit 15100 , and the entropy encoding unit 15110 may be configured by one hardware component (eg, an encoder or a processor) according to an embodiment.
  • the memory 15080 may include a decoded picture buffer (DPB), and may be configured by a digital storage medium.
  • DPB decoded picture buffer
  • the image dividing unit 15010 may divide an input image (or a picture, a frame) input to the encoding apparatus 15000 into one or more processing units.
  • the processing unit may be referred to as a coding unit (CU).
  • the coding unit may be recursively divided according to a quad-tree binary-tree (QTBT) structure from a coding tree unit (CTU) or a largest coding unit (LCU).
  • QTBT quad-tree binary-tree
  • CTU coding tree unit
  • LCU largest coding unit
  • one coding unit may be divided into a plurality of coding units having a lower depth based on a quad tree structure and/or a binary tree structure.
  • a quad tree structure may be applied first and a binary tree structure may be applied later.
  • the binary tree structure may be applied first.
  • the coding procedure according to the present invention may be performed based on the last coding unit that is no longer divided.
  • the largest coding unit may be directly used as the final coding unit based on coding efficiency according to image characteristics, or the coding unit may be recursively divided into coding units having a lower depth than the optimal coding unit if necessary.
  • a coding unit of the size of may be used as the final coding unit.
  • the coding procedure may include procedures such as prediction, transformation, and restoration, which will be described later.
  • the processing unit may further include a prediction unit (PU) or a transform unit (TU).
  • the prediction unit and the transform unit may be divided or partitioned from the above-described final coding unit, respectively.
  • the prediction unit may be a unit of sample prediction
  • the transform unit may be a unit for deriving a transform coefficient and/or a unit for deriving a residual signal from the transform coefficient.
  • a unit may be used interchangeably with terms such as a block or an area in some cases.
  • an MxN block may represent a set of samples or transform coefficients including M columns and N rows.
  • a sample may generally represent a pixel or a value of a pixel, may represent only a pixel/pixel value of a luminance component, or may represent only a pixel/pixel value of a chroma component.
  • a sample may be used as a term corresponding to a picture (or an image) as a pixel or a pel.
  • the encoding apparatus 15000 subtracts the prediction signal (predicted block, prediction sample array) output from the inter prediction unit 15090 or the intra prediction unit 15100 from the input image signal (original block, original sample array) to obtain a residual
  • a signal residual signal, residual block, residual sample array
  • a unit for subtracting a prediction signal (prediction block, prediction sample array) from an input image signal (original block, original sample array) in the encoder 15000 may be referred to as a subtraction unit 15020 .
  • the prediction unit may perform prediction on a processing target block (hereinafter, referred to as a current block) and generate a predicted block including prediction samples for the current block.
  • the prediction unit may determine whether intra prediction or inter prediction is applied on a current block or CU basis.
  • the prediction unit may generate various information related to prediction, such as prediction mode information, and transmit it to the entropy encoding unit 15110, as will be described later in the description of each prediction mode.
  • the prediction information may be encoded by the entropy encoding unit 15110 and output in the form of a bitstream.
  • the intra prediction unit 15100 may predict the current block with reference to samples in the current picture.
  • the referenced samples may be located in the vicinity of the current block according to the prediction mode, or may be located apart from each other.
  • prediction modes may include a plurality of non-directional modes and a plurality of directional modes.
  • the non-directional mode may include, for example, a DC mode and a planar mode (Planar mode).
  • the directional mode may include, for example, 33 directional prediction modes or 65 directional prediction modes according to the granularity of the prediction direction. However, this is an example, and a higher or lower number of directional prediction modes may be used according to a setting.
  • the intra prediction unit 15100 may determine the prediction mode applied to the current block by using the prediction mode applied to the neighboring block.
  • the inter prediction unit 15090 may derive the predicted block for the current block based on the reference block (reference sample array) specified by the motion vector on the reference picture.
  • the motion information may be predicted in units of blocks, subblocks, or samples based on the correlation between motion information between neighboring blocks and the current block.
  • the motion information may include a motion vector and a reference picture index.
  • the motion information may further include inter prediction direction (L0 prediction, L1 prediction, Bi prediction, etc.) information.
  • the neighboring blocks may include spatial neighboring blocks existing in the current picture and temporal neighboring blocks present in the reference picture.
  • the reference picture including the reference block and the reference picture including the temporal neighboring block may be the same or different.
  • a temporal neighboring block may be called a collocated reference block, a collocated CU (colCU), etc.
  • a reference picture including a temporal neighboring block may be called a collocated picture (colPic).
  • the inter prediction unit 15090 constructs a motion information candidate list based on neighboring blocks, and generates information indicating which candidate is used to derive a motion vector and/or a reference picture index of the current block. can do. Inter prediction may be performed based on various prediction modes. For example, in the skip mode and merge mode, the inter prediction unit 15090 may use motion information of a neighboring block as motion information of the current block.
  • the motion vector of the current block is calculated by using a motion vector of a neighboring block as a motion vector predictor and signaling a motion vector difference.
  • Inter prediction unit 15090 The prediction signal generated by the intra prediction unit 15100 may be used to generate a reconstructed signal or may be used to generate a residual signal.
  • the transform unit 15030 may generate transform coefficients by applying a transform technique to the residual signal.
  • the transformation method may include at least one of Discrete Cosine Transform (DCT), Discrete Sine Transform (DST), Karhunen-Loeve Transform (KLT), Graph-Based Transform (GBT), or Conditionally Non-linear Transform (CNT).
  • DCT Discrete Cosine Transform
  • DST Discrete Sine Transform
  • KLT Karhunen-Loeve Transform
  • GBT Graph-Based Transform
  • CNT Conditionally Non-linear Transform
  • GBT means a transformation obtained from this graph when expressing relationship information between pixels in a graph.
  • CNT refers to a transformation obtained by generating a prediction signal using all previously reconstructed pixels and based thereon.
  • the transformation process may be applied to a block of pixels having the same size as a square, or may be applied to a block of a variable size that is not a square.
  • the quantization unit 15040 quantizes the transform coefficients and transmits them to the entropy encoding unit 15110, and the entropy encoding unit 15110 encodes the quantized signal (information on the quantized transform coefficients) and outputs it as a bitstream. there is. Information about the quantized transform coefficients may be referred to as residual information.
  • the quantization unit 15040 may rearrange the quantized transform coefficients in a block form into a one-dimensional vector form based on a coefficient scan order, and a quantized transform coefficient based on the quantized transform coefficients in a one-dimensional vector form. You can also create information about them.
  • the entropy encoding unit 15110 may perform various encoding methods such as, for example, exponential Golomb, context-adaptive variable length coding (CAVLC), and context-adaptive binary arithmetic coding (CABAC).
  • the entropy encoding unit 15110 may encode information necessary for video/image reconstruction (eg, values of syntax elements, etc.) other than the quantized transform coefficients together or separately.
  • Encoded information eg, encoded video/image information
  • NAL network abstraction layer
  • the bitstream may be transmitted over a network, or may be stored in a digital storage medium.
  • the network may include a broadcasting network and/or a communication network
  • the digital storage medium may include various storage media such as USB, SD, CD, DVD, Blu-ray, HDD, and SSD.
  • the transmitting unit (not shown) and/or the storing unit (not shown) for storing the signal may be configured as internal/external elements of the encoding apparatus 15000, or the transmitting unit It may be included in the entropy encoding unit 15110 .
  • the quantized transform coefficients output from the quantization unit 15040 may be used to generate a prediction signal.
  • the residual signal residual block or residual samples
  • the adder 155 adds the reconstructed residual signal to the prediction signal output from the inter prediction unit 15090 or the intra prediction unit 15100 to obtain a reconstructed signal (reconstructed picture, reconstructed block, reconstructed sample array). can be created
  • the predicted block may be used as a reconstructed block.
  • the adder 155 may be referred to as a restoration unit or a restoration block generator.
  • the generated reconstructed signal may be used for intra prediction of the next processing target block in the current picture, or may be used for inter prediction of the next picture after filtering as described below.
  • the filtering unit 15070 may improve subjective/objective image quality by applying filtering to the reconstructed signal. For example, the filtering unit 15070 may generate a modified reconstructed picture by applying various filtering methods to the reconstructed picture, and the modified reconstructed picture is stored in the memory 15080, specifically, in the DPB of the memory 15080. can be saved Various filtering methods may include, for example, deblocking filtering, a sample adaptive offset, an adaptive loop filter, a bilateral filter, and the like. The filtering unit 15070 may generate various information regarding filtering and transmit it to the entropy encoding unit 15110, as will be described later in the description of each filtering method. The filtering-related information may be encoded by the entropy encoding unit 15110 and output in the form of a bitstream.
  • the modified reconstructed picture transmitted to the memory 15080 may be used as a reference picture in the inter prediction unit 15090 .
  • the encoding apparatus can avoid prediction mismatch between the encoding apparatus 15000 and the decoding apparatus, and can also improve encoding efficiency.
  • the memory 15080 DPB may store the modified reconstructed picture to be used as a reference picture in the inter prediction unit 15090 .
  • the memory 15080 may store motion information of a block in which motion information in the current picture is derived (or encoded) and/or motion information of blocks in an already reconstructed picture.
  • the stored motion information may be transmitted to the inter prediction unit 15090 to be used as motion information of a spatial neighboring block or motion information of a temporal neighboring block.
  • the memory 15080 may store reconstructed samples of blocks reconstructed in the current picture, and may transmit the reconstructed samples to the intra prediction unit 15100 .
  • prediction, transformation, and quantization procedures may be omitted.
  • prediction, transformation, and quantization procedures may be omitted, and the value of the original sample may be encoded and output as a bitstream.
  • V-PCC decoding process shows an example of a V-PCC decoding process according to embodiments.
  • V-PCC decoding process or V-PCC decoder may follow the reverse process of the V-PCC encoding process (or encoder) of FIG. 4 .
  • Each component in FIG. 16 may correspond to software, hardware, a processor, and/or a combination thereof.
  • a demultiplexer demultiplexes the compressed bitstream to output a compressed texture image, a compressed geometry image, a compressed occupanci map, and compressed accessory patch information.
  • the video decompression (video decompression, 16001 and 16002) or video decompressor decompresses (or decodes) each of a compressed texture image and a compressed geometry image.
  • An occupancy map decompression (16003) or occupancy map decompressor decompresses a compressed occupancy map.
  • auxiliary patch info decompression 16004
  • auxiliary patch information decompressor decompresses auxiliary patch information.
  • a geometry reconstruction (16005) or geometry reconstructor reconstructs (reconstructs) geometry information based on a decompressed geometry image, a decompressed accumulatory map, and/or decompressed accupancy patch information. For example, a geometry changed in the encoding process may be reconstructed.
  • a smoothing (16006) or smoother may apply smoothing to the reconstructed geometry. For example, smoothing filtering may be applied.
  • a texture reconstruction (16007) or texture reconstructor reconstructs a texture from a decompressed texture image and/or smoothed geometry.
  • a color smoothing (16008) or color smoother smoothes color values from the reconstructed texture. For example, smoothing filtering may be applied.
  • reconstructed point cloud data may be generated.
  • the figure shows and shows the decoding process of V-PCC for reconstructing the point cloud by decoding the compressed occupancy map, geometry image, texture image, and auxiliary path information. same. Operations of each process according to the embodiments are as follows.
  • a reverse process of video compression described above is a process of decoding a compressed bitstream such as a geometry image, texture image, and occupancy map image generated by the process described above using a 2D video codec such as HEVC and VVC.
  • a 2D video codec such as HEVC and VVC.
  • FIG 17 shows an example of a 2D video/image decoder according to embodiments.
  • the 2D video/image decoder may follow the reverse process of the 2D video/image encoder of FIG. 15 .
  • the 2D video/image decoder of FIG. 17 is an embodiment of the video decompression or video decompressor of FIG. 16, and is a schematic block diagram of a 2D video/image decoder 17000 in which decoding of a video/image signal is performed. indicates
  • the 2D video/image decoder 17000 may be included in the point cloud video decoder of FIG. 1 , or may be composed of internal/external components. Each component of FIG. 17 may correspond to software, hardware, a processor, and/or a combination thereof.
  • the input bitstream may include a bitstream for the above-described geometry image, texture image (attribute(s) image), occupancy map image, and the like.
  • the reconstructed image (or output image, decoded image) may represent a reconstructed image for the above-described geometry image, texture image (attribute(s) image), and occupancy map image.
  • the inter prediction unit 17070 and the intra prediction unit 17080 may be collectively referred to as a prediction unit. That is, the prediction unit may include an inter prediction unit 180 and an intra prediction unit 185 .
  • the inverse quantization unit 17020 and the inverse transform unit 17030 may be collectively referred to as a residual processing unit. That is, the residual processing unit may include an inverse quantization unit 17020 and an inverse transform unit 17030 .
  • the above-described entropy decoding unit 17010, inverse quantization unit 17020, inverse transform unit 17030, adder 17040, filtering unit 17050, inter prediction unit 17070 and intra prediction unit 17080 are the embodiment may be configured by one hardware component (eg, a decoder or a processor).
  • the memory 170 may include a decoded picture buffer (DPB), and may be configured by a digital storage medium.
  • DPB decoded picture buffer
  • the decoding apparatus 17000 may reconstruct an image corresponding to a process in which the video/image information is processed in the encoding apparatus of FIG. 0.2-1 .
  • the decoding apparatus 17000 may perform decoding using a processing unit applied in the encoding apparatus.
  • the processing unit of decoding may be, for example, a coding unit, and the coding unit may be divided along a quad tree structure and/or a binary tree structure from a coding tree unit or a largest coding unit.
  • the restored image signal decoded and output through the decoding device 17000 may be reproduced through the playback device.
  • the decoding apparatus 17000 may receive a signal output from the encoding apparatus in the form of a bitstream, and the received signal may be decoded through the entropy decoding unit 17010 .
  • the entropy decoding unit 17010 may parse the bitstream to derive information (eg, video/image information) required for image restoration (or picture restoration).
  • the entropy decoding unit 17010 decodes information in a bitstream based on a coding method such as exponential Golomb encoding, CAVLC or CABAC, and a value of a syntax element required for image reconstruction, and a quantized value of a transform coefficient related to a residual can be printed out.
  • the CABAC entropy decoding method receives a bin corresponding to each syntax element in the bitstream, and decodes the syntax element information to be decoded and the decoding information of the surrounding and decoding target blocks or the symbol/bin information decoded in the previous step.
  • a context model is determined using the context model, and the probability of occurrence of a bin is predicted according to the determined context model, and a symbol corresponding to the value of each syntax element can be generated by performing arithmetic decoding of the bin.
  • the CABAC entropy decoding method may update the context model by using the decoded symbol/bin information for the context model of the next symbol/bin after determining the context model.
  • Prediction-related information among the information decoded by the entropy decoding unit 17010 is provided to the prediction unit (the inter prediction unit 17070 and the intra prediction unit 265), and the entropy decoding unit 17010 performs entropy decoding.
  • the dual value, that is, the quantized transform coefficients and related parameter information may be input to the inverse quantization unit 17020 .
  • information about filtering among the information decoded by the entropy decoding unit 17010 may be provided to the filtering unit 17050 .
  • a receiving unit (not shown) that receives a signal output from the encoding device may be further configured as an internal/external element of the decoding device 17000 , or the receiving unit may be a component of the entropy decoding unit 17010 .
  • the inverse quantizer 17020 may inverse quantize the quantized transform coefficients to output the transform coefficients.
  • the inverse quantizer 17020 may rearrange the quantized transform coefficients in a two-dimensional block form. In this case, the rearrangement may be performed based on the coefficient scan order performed by the encoding device.
  • the inverse quantizer 17020 may perform inverse quantization on the quantized transform coefficients using a quantization parameter (eg, quantization step size information) and obtain transform coefficients.
  • a quantization parameter eg, quantization step size information
  • the inverse transform unit 17030 inverse transforms the transform coefficients to obtain a residual signal (residual block, residual sample array).
  • the prediction unit may perform prediction on the current block and generate a predicted block including prediction samples for the current block.
  • the prediction unit may determine whether intra prediction or inter prediction is applied to the current block based on the prediction information output from the entropy decoding unit 17010, and may determine a specific intra/inter prediction mode.
  • the intra prediction unit 265 may predict the current block with reference to samples in the current picture.
  • the referenced samples may be located in the vicinity of the current block according to the prediction mode, or may be located apart from each other.
  • prediction modes may include a plurality of non-directional modes and a plurality of directional modes.
  • the intra prediction unit 265 may determine the prediction mode applied to the current block by using the prediction mode applied to the neighboring block.
  • the inter prediction unit 17070 may derive the predicted block for the current block based on the reference block (reference sample array) specified by the motion vector on the reference picture.
  • the motion information may be predicted in units of blocks, subblocks, or samples based on the correlation between motion information between neighboring blocks and the current block.
  • the motion information may include a motion vector and a reference picture index.
  • the motion information may further include inter prediction direction (L0 prediction, L1 prediction, Bi prediction, etc.) information.
  • the neighboring blocks may include spatial neighboring blocks existing in the current picture and temporal neighboring blocks present in the reference picture.
  • the inter prediction unit 17070 may construct a motion information candidate list based on neighboring blocks, and derive a motion vector and/or a reference picture index of the current block based on the received candidate selection information. Inter prediction may be performed based on various prediction modes, and the prediction information may include information indicating the inter prediction mode for the current block.
  • the adder 17040 adds the obtained residual signal to the prediction signal (predicted block, prediction sample array) output from the inter prediction unit 17070 or the intra prediction unit 265 to obtain a reconstructed signal (reconstructed picture, reconstructed block). , a reconstructed sample array) can be created.
  • the predicted block may be used as a reconstructed block.
  • the addition unit 17040 may be called a restoration unit or a restoration block generation unit.
  • the generated reconstructed signal may be used for intra prediction of the next processing object block in the current picture, or may be used for inter prediction of the next picture after filtering as described below.
  • the filtering unit 17050 may improve subjective/objective image quality by applying filtering to the reconstructed signal.
  • the filtering unit 17050 may generate a modified reconstructed picture by applying various filtering methods to the reconstructed picture, and stores the modified reconstructed picture in the memory 17060, specifically, in the DPB of the memory 17060.
  • Various filtering methods may include, for example, deblocking filtering, a sample adaptive offset, an adaptive loop filter, a bilateral filter, and the like.
  • the (modified) reconstructed picture stored in the DPB of the memory 17060 may be used as a reference picture in the inter prediction unit 17070 .
  • the memory 17060 may store motion information of a block in which motion information in the current picture is derived (or decoded) and/or motion information of blocks in an already reconstructed picture.
  • the stored motion information may be transmitted to the inter prediction unit 17070 to be used as motion information of a spatial neighboring block or motion information of a temporal neighboring block.
  • the memory 170 may store reconstructed samples of blocks reconstructed in the current picture, and may transmit the reconstructed samples to the intra prediction unit 17080 .
  • the embodiments described in the filtering unit 160, the inter prediction unit 180, and the intra prediction unit 185 of the encoding apparatus 100 are the filtering unit 17050 and the inter prediction unit of the decoding apparatus 17000, respectively.
  • the same or corresponding application may be applied to the unit 17070 and the intra prediction unit 17080 .
  • prediction, transformation, and quantization procedures may be omitted.
  • prediction, transformation, and quantization procedures may be omitted, and a value of a decoded sample may be used as a sample of a reconstructed image as it is.
  • This is the reverse process of the occupancy map compression described above, and is a process for restoring the occupancy map by decoding the compressed occupancy map bitstream.
  • Auxiliary patch info can be restored by performing the reverse process of auxiliary patch info compression described above and decoding the compressed auxiliary patch info bitstream.
  • the patch is extracted from the geometry image using the 2D location/size information of the patch and the mapping information of the block and the patch included in the restored occupancy map and auxiliary patch info.
  • the point cloud is restored in 3D space using the extracted patch geometry image and the patch 3D location information included in auxiliary patch info.
  • the color values corresponding to the texture image pixels in the same position as in the geometry image in 2D space are obtained from the point cloud corresponding to the same position in the 3D space. It can be done by giving a point.
  • smoothing may be performed by determining a portion other than an edge.
  • a method of smoothing a method of changing the color value of a corresponding point with the average value of adjacent tangents may be used.
  • FIG. 18 shows an example of an operation flowchart of a transmitting apparatus according to the embodiments.
  • the transmitting apparatus may correspond to the transmitting apparatus of FIG. 1 , the encoding process of FIG. 4 , and the 2D video/image encoder of FIG. 15 or perform some/all operations thereof.
  • Each component of the transmitting device may correspond to software, hardware, a processor and/or a combination thereof.
  • An operation process of the transmitter for compression and transmission of point cloud data using V-PCC may be as shown in the drawing.
  • the point cloud data transmission apparatus may be referred to as a transmission apparatus or the like.
  • a patch for mapping a 2D image of a point cloud is generated. Additional patch information is generated as a result of patch generation, and the corresponding information can be used for geometry image generation, texture image generation, and geometry restoration for smoothing.
  • the generated patches are subjected to a patch packing process for mapping into a 2D image.
  • a patch packing process for mapping into a 2D image.
  • an occupancy map can be generated, and the occupancy map can be used for geometry image generation, texture image generation, and geometry restoration for smoothing.
  • the geometry image generator 18002 generates a geometry image by using the additional patch information and the occupanci map, and the generated geometry image is encoded into one bitstream through video encoding.
  • the encoding preprocessing 18003 may include an image padding procedure.
  • the generated geometry image or the geometry image regenerated by decoding the encoded geometry bitstream may be used for 3D geometry restoration and may then be subjected to a smoothing process.
  • the texture image generator 18004 may generate a texture image by using a (smoothed) 3D geometry, a point cloud, additional patch information, and an occupanci map.
  • the generated texture image may be encoded into one video bitstream.
  • the metadata encoder 18005 may encode the additional patch information into one metadata bitstream.
  • the video encoder 18006 may encode the occult map into one video bitstream.
  • the multiplexer 18007 multiplexes a video bitstream of the generated geometry, texture image, and occupancy map and an additional patch information metadata bitstream into one bitstream.
  • the transmitter 18008 may transmit the bitstream to the receiver.
  • the video bitstream of the generated geometry, texture image, occupanci map, and the additional patch information metadata bitstream may be created as one or more track data files or encapsulated into segments and transmitted to the receiver through the transmitter.
  • FIG. 19 shows an example of an operation flowchart of a receiving apparatus according to the embodiments.
  • the receiving device may correspond to the receiving device of FIG. 1 , the decoding process of FIG. 16 , and the 2D video/image encoder of FIG. 17 or perform some/all operations thereof.
  • Each component of the receiving device may correspond to software, hardware, a processor and/or a combination thereof.
  • An operation process of the receiving end for receiving and restoring point cloud data using V-PCC may be as shown in the drawing.
  • the operation of the V-PCC receiving end may follow the reverse process of the operation of the V-PCC transmitting end of FIG. 18 .
  • a device for receiving point cloud data may be referred to as a receiving device or the like.
  • the received bitstream of the point cloud is demultiplexed by the demultiplexer 19000 into an additional patch information metadata bitstream and video bitstreams of a compressed geometry image, texture image, occupanci map after file/segment decapsulation. do.
  • the video decoder 19001 and the metadata decoder 19002 decode demultiplexed video bitstreams and metadata bitstreams.
  • the 3D geometry is restored by using the geometry image decoded by the geometry restoration unit 19003, the occupancy map, and additional patch information, and then undergoes a smoothing process by the smoother 19004.
  • the color point cloud image/picture may be reconstructed by the texture restoration unit 19005 by giving a color value to the smoothed 3D geometry using a texture image.
  • a color smoothing process can be additionally performed to improve the objective/subjective visual quality, and the modified point cloud image/picture derived through this can be processed through the rendering process (ex. by point cloud renderer) displayed to the user through Meanwhile, the color smoothing process may be omitted in some cases.
  • FIG. 20 shows an example of a structure capable of interworking with a method/device for transmitting and receiving point cloud data according to embodiments.
  • a structure according to embodiments may include at least one of a server 2360 , a robot 2010 , an autonomous vehicle 2020 , an XR device 2030 , a smartphone 2040 , a home appliance 2050 , and/or an HMD 2070 .
  • the above is connected to the cloud network 2010 .
  • the robot 2010 , the autonomous driving vehicle 2020 , the XR device 2030 , the smartphone 2040 , or the home appliance 2050 may be referred to as devices.
  • the XR device 2030 may correspond to a point cloud data (PCC) device according to embodiments or may be linked with the PCC device.
  • PCC point cloud data
  • the cloud network 2000 may refer to a network that constitutes a part of the cloud computing infrastructure or exists in the cloud computing infrastructure.
  • the cloud network 2000 may be configured using a 3G network, a 4G or Long Term Evolution (LTE) network, or a 5G network.
  • LTE Long Term Evolution
  • the server 2360 includes at least one of a robot 2010, an autonomous vehicle 2020, an XR device 2030, a smartphone 2040, a home appliance 2050, and/or an HMD 2070, and a cloud network 2000. It is connected through and may help at least a part of the processing of the connected devices 2010 to 2070 .
  • a Head-Mount Display (HMD) 2070 represents one of the types in which an XR device and/or a PCC device according to embodiments may be implemented.
  • the HMD type device according to the embodiments includes a communication unit, a control unit, a memory unit, an I/O unit, a sensor unit, a power supply unit, and the like.
  • the devices 2010 to 2070 to which the above-described technology is applied will be described.
  • the devices 2000 to 2700 shown in FIG. 20 may be linked/coupled with the point cloud data transmission/reception device according to the above-described embodiments.
  • the XR/PCC device 2030 is a PCC and/or XR (AR+VR) technology applied, such as a Head-Mount Display (HMD), a Head-Up Display (HUD) provided in a vehicle, a television, It may be implemented as a mobile phone, a smart phone, a computer, a wearable device, a home appliance, a digital signage, a vehicle, a stationary robot, or a mobile robot.
  • HMD Head-Mount Display
  • HUD Head-Up Display
  • the XR/PCC device 2030 analyzes three-dimensional point cloud data or image data acquired through various sensors or from an external device to generate position data and attribute data for three-dimensional points in the surrounding space or real objects. Information can be obtained and the XR object to be output can be rendered and output. For example, the XR/PCC apparatus 2030 may output an XR object including additional information on the recognized object to correspond to the recognized object.
  • the autonomous driving vehicle 2020 may be implemented as a mobile robot, vehicle, or unmanned aerial vehicle by applying PCC technology and XR technology.
  • the autonomous driving vehicle 2020 to which the XR/PCC technology is applied may mean an autonomous driving vehicle equipped with a means for providing an XR image, an autonomous driving vehicle subject to control/interaction within the XR image, or the like.
  • the autonomous driving vehicle 2020 that is the target of control/interaction within the XR image may be distinguished from the XR device 2030 and may be interlocked with each other.
  • the autonomous vehicle 2020 having means for providing an XR/PCC image may obtain sensor information from sensors including a camera, and output an XR/PCC image generated based on the acquired sensor information.
  • the autonomous vehicle may provide the occupant with an XR/PCC object corresponding to a real object or an object in a screen by having a HUD and outputting an XR/PCC image.
  • the XR/PCC object when the XR/PCC object is output to the HUD, at least a portion of the XR/PCC object may be output to overlap the real object to which the passenger's gaze is directed.
  • the XR/PCC object when the XR/PCC object is output to a display provided inside the autonomous vehicle, at least a portion of the XR/PCC object may be output to overlap the object in the screen.
  • the autonomous vehicle may output XR/PCC objects corresponding to objects such as a lane, other vehicles, traffic lights, traffic signs, two-wheeled vehicles, pedestrians, and buildings.
  • VR Virtual Reality
  • AR Augmented Reality
  • MR Magnetic Reality
  • PCC Point Cloud Compression
  • VR technology is a display technology that provides objects or backgrounds in the real world only as CG images.
  • AR technology refers to a technology that shows a virtual CG image on top of a real object image.
  • the MR technology is similar to the AR technology described above in that it shows the virtual objects by mixing and combining them in the real world.
  • AR technology the distinction between real objects and virtual objects made of CG images is clear, and virtual objects are used in a form that complements real objects, whereas in MR technology, virtual objects are regarded as having the same characteristics as real objects. distinct from technology. More specifically, for example, a hologram service to which the aforementioned MR technology is applied.
  • VR, AR, and MR technologies are sometimes called XR (extended reality) technologies rather than clearly distinguishing them. Accordingly, embodiments of the present invention are applicable to all of VR, AR, MR, and XR technologies. As one such technology, encoding/decoding based on PCC, V-PCC, and G-PCC technology may be applied.
  • the PCC method/apparatus according to the embodiments may be applied to a vehicle providing an autonomous driving service.
  • a vehicle providing an autonomous driving service is connected to a PCC device to enable wired/wireless communication.
  • the point cloud data (PCC) transceiver receives/processes AR/VR/PCC service-related content data that can be provided together with the autonomous driving service when connected to a vehicle to enable wired/wireless communication, can be sent to
  • the point cloud transceiver may receive/process AR/VR/PCC service-related content data according to a user input signal input through the user interface device and provide it to the user.
  • a vehicle or a user interface device may receive a user input signal.
  • a user input signal according to embodiments may include a signal indicating an autonomous driving service.
  • FIG. 1 transmitting apparatus 10000 point cloud video encoder 10002, FIG. 4 encoding process, FIG. 15 video/image encoder, FIG. 18 transmitting apparatus, and FIG. 20 XR device 1730, and may refer to a transmission apparatus in FIG. 40 and the like.
  • Each component of the transmission method/apparatus may correspond to hardware, software, a processor coupled with a memory, and/or a combination thereof.
  • a method/apparatus for receiving point cloud data is shown in Fig. 2 receiving apparatus 10005, point cloud video decoder 10008, Fig. 16 decoding process, Fig. 17 video/image decoder, Fig. 19 receiving apparatus, and Fig. 20 XR device. (1730) and the like.
  • Each component of the receiving method/apparatus may correspond to hardware, software, a processor coupled with memory, and/or a combination thereof.
  • a method/device for transmitting and receiving point cloud data according to embodiments may be abbreviated as a method/device according to embodiments.
  • FIG. 21 illustrates a voxelized surface light field sequence according to embodiments.
  • the method/apparatus according to the embodiments may compress and restore a Surface Light Field (SLF) data set including point cloud data.
  • SLF Surface Light Field
  • the method/apparatus according to the embodiments may use a patch-based camera view selection and attribute video generation method for V-PCC-based effective Surface Light Field (SLF) sequence compression/restore.
  • SLF Surface Light Field
  • Embodiments relate to Video-based Point Cloud Compression (V-PCC), which is a method of compressing 3D point cloud data using a 2D video codec.
  • V-PCC Video-based Point Cloud Compression
  • SLF Surface Light Field
  • the embodiments provide a method for selecting a significant camera viewpoint among camera viewpoints constituting SLF sequence data in the V-PCC, a method for generating an attribute video stream, a method for selecting a camera view, and a method for generating a video stream. ), and video stream processing methods in the transmitter and receiver.
  • the V-PCC method uses three video streams from the input data to compress the input three-dimensional point cloud data, 1) Occupancy map video, 2) Geometry video, 3) Attribute ) to create a video stream. And they are each compressed using the 2D video codec in the V-PCC encoder. Among them, the attribute video is generated to include the color attribute of the 3D point on the 2D image. If the attribute video has texture type information, each pixel value on the image is the ( R, G, B) Indicates the color attribute.
  • the Voxelized Surface Light Field (VSLF) data set provides a more realistic experience with more realistic expression and free movement of the viewpoint when rendering 3D scenes in applications such as real and virtual reality. It is a test data set for standardization of MPEG made to provide users with (see FIG. 21). Unlike the general point cloud data set, which has one (R, G, B) information as a color property representing one point, this SLF data set has as many (R, G, B) as the number of different camera viewpoints. have information Therefore, when compressing a general point cloud data set using V-PCC, one attribute video can be created to express and compress each color attribute of all points, but when compressing an SLF data set, one attribute video is created. Since raw data cannot contain all color properties, additional V-PCC compression technology must be applied in consideration of these characteristics.
  • Embodiments can generate and compress multiple attribute videos by applying the same method of generating the color attribute of the point cloud as one attribute video in the V-PCC compression process to the color attributes of all SLF data sets.
  • the embodiments In the case of compressing the sequence of FIG. 21, the embodiments generate one geometry video, one occupancy map, and 13 attribute videos along with auxiliary information. And, in the same way that V-PCC compresses each video stream using a 2D video codec, each of the 13 attribute videos is compressed.
  • the compressed data when the same sequence was compressed in the form of general point cloud data rather than SLF data (2 frames, highest quality compression condition), the compressed data was 991099 Bytes to 0.9 Bytes, and the compressed data when compressed in the form of SLF data set was It may have a capacity of 11083828 Bytes to 11 Mega bytes. Since the original data size of the SLF sequence is large, the data size of the compressed result is also very large. When this is applied to an actual application, it can be predicted that a high level of memory capacity will be required. In addition, there is a disadvantage that a large number of video codec instances are required to encode and decode 13 attribute videos, respectively.
  • FIG. 22 shows an example in which 2D point cloud data according to embodiments are displayed separately for each point according to activity region information.
  • Embodiments may include more optimized techniques to solve this excessive memory usage problem, in other words a low compression ratio problem.
  • a representative camera that can represent all camera viewpoints is selected, and only the color information expressed by these camera views is compressed.
  • all points of the SLF sequence may be classified into a low (low, 2200), a medium (medium, 2201), and a high activity region (2202) (refer to FIG. 22).
  • the method/apparatus according to the embodiments performs patch segmentation in consideration of activity region information. Through this process, all points in one patch have the same activity region, and 5 representative cameras are selected for a patch having a high activity region, and a medium activity region is selected. Three representative cameras may be selected for a patch having a medium activity region, and one representative camera may be selected for a patch having a low activity region. An attribute video is created using the color information expressed by the camera selected in this way.
  • FIG. 23 shows a configuration of a 5-attribute video and 1-geometry video generated from an SLF data set according to embodiments.
  • the method/apparatus according to the embodiments may solve the problem of the memory requirement of the described method, and may further provide [additional operations for providing high-quality compression performance.
  • a method for effectively compressing color data of an SLF sequence using V-PCC which does not compress color information of all camera viewpoints for each point, and selects only meaningful color information of camera viewpoints in units of patches
  • Method method of generating and compressing video using color information of selected camera viewpoints, method of minimizing encoding and decoding complexity and the number of required codec instances by minimizing the number of attribute videos used in V-PCC, transmitting end (or V -PCC encoder) and the receiving end (or V-PCC decoder) will explain the signaling method to let you know that the SLF sequence compression method has been applied.
  • Attribute video refers to a video generated using attribute data (or attributes) supported by V-PCC, and if the corresponding attribute video has a texture data type may refer to this as a texture video.
  • the SLF sequence means data composed of color information obtained from multiple cameras for one point. That is, the SLF sequence includes several pieces of color information as data for each point of the point cloud representing the object. Several pieces of color information for one point may have different values depending on a photographing position and angle of the camera. Embodiments provide a method of selecting only meaningful color information in units of patches when generating a texture video with respect to color information of an SLF sequence, and generating a video using them.
  • Embodiments use two methods for selecting a camera viewpoint for each patch.
  • the total number of camera viewpoints constituting the SLF data is defined as N, and the number of camera viewpoints selected for each patch is defined as M.
  • FIG. 24 shows an example of a method for selecting a camera viewpoint for each patch and generating a texture video according to embodiments.
  • the patch packing step and the patch packer (processor) of the point cloud data transmission method/device according to the embodiments may perform this operation, and this operation in the patch decoding process of the point cloud data receiving method/device according to the embodiments
  • the patch can be decoded based on
  • Embodiments select M camera viewpoints closest to the direction in which the patch is facing from among the N camera viewpoints constituting the SLF data (refer to FIG. 24).
  • the point area that the user can actually see will most likely correspond to the front part of each area. Therefore, even if there are N pieces of color information for one point, information that an actual user can see and need is highly likely to be limited to color information corresponding to some angles.
  • the embodiments determine that the color information obtained from the camera viewpoint existing in the front direction of the patch is the most effective and accurate based on these characteristics, so that only the corresponding information can be selected, compressed, and transmitted.
  • Each of the N camera viewpoints used to generate SLF data has camera matrix information for its position and direction.
  • the projected direction of the corresponding patch is determined, so that the direction the patch faces in a 3D space can be known through a normal vector. Therefore, it is possible to determine the camera viewpoint closest to the patch through the operation between the camera matrix information and the normal vector of the patch.
  • the operation between the camera matrix and the normal vector of the patch is the vector dot product (u v), and in this case, the closer the patch and the camera viewpoint to each other, the smaller the result of the vector dot product operation. Therefore, if the vector dot product operation between the camera matrix of all camera viewpoints and the patch normal vector is performed, and M camera viewpoints having the smallest value are selected among them, the camera viewpoints closest to the patch can be selected. there is.
  • the M value can be set differently for each patch, and in this case, an additional constraint can be applied to exclude the case where the angle between the camera matrix direction vector and the patch normal vector is out of a certain angle range.
  • the tile may have the same M value, or the frame may have the same M value. Similarly, it may be set to have the same M value for the same sequence.
  • Information on the setting of the M value is signaled as atlas information.
  • information on the M number of selected camera viewpoints is signaled for each patch and transmitted. As the value of M increases, the number of generated texture videos increases, but a view coverage range for a patch may be widened.
  • the texture video may have number 0, number 1, number 2, and the like.
  • Each texture video includes patch(s) corresponding to the texture video, and the patches may be obtained from the camera(s) closest to the patch and included in the texture video.
  • texture video #0 may include patches obtained from camera #4, camera #6, and camera #1. That is, texture video #0 includes the closest camera #4 based patch among camera #5, camera #4, and camera #3. In this way, it is possible to configure the point cloud data for the object in detail, and even if the amount of data increases, it can be efficiently compressed.
  • the method/device according to the embodiments may exclude overlapping attributes (colors) between cameras in a patch.
  • the SLF data for each point contains N pieces of color information, and there are many cases where it has overlapping color information depending on the position and direction of the point and the camera viewpoint. Therefore, it may be a more efficient method to select and transmit only representative color information excluding duplicate color information.
  • Embodiments determine and select a camera viewpoint that does not overlap with each other and has more differentiated color information among color information of camera viewpoints expressing a patch as a meaningful camera viewpoint.
  • the camera viewpoint closest to the direction of the patch is selected by using the camera matrix of the used camera viewpoint and the operation of the normal vector of the patch. This is defined as the representative camera viewpoint cam_rep that best expresses the color information of the patch. Then, the difference between the patch color obtained from the remaining N-1 camera viewpoints and the patch color obtained from cam_rep is calculated, respectively.
  • a camera viewpoint other than cam_rep is defined as cam_k, and the difference between the color values obtained from cam_rep and cam_k is calculated as 2400 for all points constituting the patch.
  • ave_C_diff_k After calculating the 2401 values of all points in the patch, their average value ave_C_diff_k is calculated. Ave_C_diff is calculated for the remaining N - 1 camera viewpoints except for cam_rep, and M camera viewpoints are selected in the order of the largest value. As ave_C_diff is larger, it can be determined that colors having a large difference from the color obtained from cam_rep can be obtained from the corresponding camera viewpoint, and can be meaningful information as non-overlapping differential color information. As an additional constraint, transmission of similar color information can be excluded by not selecting a camera viewpoint whose ave_C_diff is less than or equal to a certain threshold.
  • one method can be used and applied according to the user's selection, and the used selection method is signaled so that the receiving end can know.
  • the user at the receiving end can know how the corresponding SLF sequence color data is selected and transmitted, and can selectively utilize the compressed SLF color information according to the purpose of use.
  • the order of mapping color information for each patch to each of the K texture videos reflects the order in which camera viewpoints are selected for each patch. When the camera viewpoint selection method 1 for each patch is used, the colors of the camera viewpoints are mapped to each texture video in the order in which the direction toward the patch is closest (refer to Fig. 24).
  • 2402 is a patch having a red color and camera views for acquiring the patch
  • 2403 is a patch having a blue color and camera views for acquiring the patch
  • 2404 is a patch having a yellow color and camera views thereof These are the camera views that get the patch.
  • the color information obtained from cam_rep is mapped to the first texture video, and from then on, the color information obtained from the corresponding camera viewpoint is mapped to the texture video in the order of increasing ave_C_diff. .
  • Applying this order is to reflect the importance of color information in the order of the texture video, and accordingly, it can be considered that the first texture video contains the most meaningful color information.
  • application of this order may help the receiving end to determine whether to restore all colors or only some color data having high importance according to a user's selection.
  • a patch without color information to be mapped may occur in some cases.
  • the patch color in the corresponding texture video is filled with 0.
  • the K texture videos generated in this way are respectively compressed and transmitted using a 2D video codec.
  • 25 illustrates a texture video generation method according to embodiments.
  • texture video creation method 1 (2500) As the number of texture videos created increases, more codec instances are required. In this texture video creation method 2, the number of texture videos created is reduced.
  • K texture videos are created in the same way as in method 1 for creating texture videos, and then, each frame of each video is combined into one frame and combined into one texture video 2501 (refer to FIG. 25).
  • the number of video frames arranged in the horizontal/vertical direction may be determined according to a user's designation.
  • the information of each frame is arranged in the raster scan order in the synthesized texture video. If the frame area that cannot be filled in the synthesized texture video is filled with 0.
  • the synthesized texture video is compressed and transmitted using one video codec.
  • the attribute video transmitted through the signaling information (parameter information) is a synthesized video, and the receiving end knows about the structure of the synthesized video by signaling the number of videos existing in the horizontal/vertical direction make it possible
  • 26 shows a V3C bitstream structure according to embodiments.
  • a method/apparatus for transmitting point cloud data may compress (encode) point cloud data, generate related parameter information (eg, FIGS. 26 to 39 ), and may generate and transmit a bitstream as shown in FIG. 26 .
  • a method/apparatus for receiving point cloud data may receive a bitstream as shown in FIG. 26 and decode the point cloud data included in the bitstream based on parameter information included in the bitstream.
  • Signaling information (which may be referred to as a parameter/metadata, etc.) according to the embodiments is encoded by a metadata incubator (which may be referred to as a metadata encoder, etc.) in the point cloud data transmission apparatus according to the embodiments to be transmitted to a bitstream may be included and transmitted. Also, in the point cloud data receiving apparatus according to the embodiments, it may be decoded by a metadata decoder (which may be referred to as a metadata decoder, etc.) and provided to a decoding process of the point cloud data.
  • a metadata incubator which may be referred to as a metadata encoder, etc.
  • a transmitter may generate a bitstream by encoding the point cloud data.
  • a bitstream according to embodiments may include a V3C unit.
  • a receiver may receive a bitstream transmitted by a transmitter, and may decode and restore point cloud data.
  • V3C unit may decode and restore point cloud data.
  • V3C parameter set syntax generated by a method/apparatus for transmitting point cloud data according to embodiments.
  • Parameter set ID (vps_v3c_parameter_set_id): an identifier for the VPS for reference by other syntax elements.
  • vps_reserved_zero_8bits may be 0 in the bitstream.
  • Atlas count (vps_atlas_count_minus1): If 1 is added to this value, the total number of supported atlases in the current bitstream. It may have a value of 0 to 63.
  • Atlas ID (vps_atlas_id[ k ]): Atlas ID with index K. It may have a value of 0 to 63.
  • Map count (vps_map_count_minus1[ j ]): If 1 is added to this value, it indicates the number of maps used to encode the geometry and attribute data for the atlas with atlas ID J. It may have a value of 0 to 15.
  • vps_multiple_map_streams_present_flag[ j ] If this value is 0, it indicates that all geometry or attribute maps for the atlas with atlas IDJ are located in a single geometry or attribute video stream. If this value is 1, it indicates that all geometry or attribute maps for the atlas with atlas idJ are present in individual video streams.
  • Map absolute coding enable flag (vps_map_absolute_coding_enabled_flag[ j ][ i ]): If this value is 1, it indicates that the geometry map with index I for the atlas with atlas ID J is coded without map prediction. If this value is 0, it indicates that the geometry map with index I for the atlas with atlas id J is predicted first from others coded before map coding.
  • Map predictor index difference (vps_map_predictor_index_diff[ j ][ i ]): If vps_map_absolute_coding_enabled_flag[ j ][ i ] is 0, it is used to compute the predictor of the geometry map with index I for the atlas with atlas id J.
  • Auxiliary video presence flag (vps_auxiliary_video_present_flag[ j ]): If this value is 1, additional information about the patch in the atlas with atlas ID J, that is, RAW or EOM patch types related information, can be stored in an individual video stream. indicates. The individual video streams may be referred to as oscillary video streams. If this value is 0, it indicates that additional information about the patch in the atlas having the atlas ID J, ie, information related to RAW or EOM patch types, is not stored in the oscillation video stream.
  • Geometry video presence flag (vps_geometry_video_present_flag[ j ]): If this value is 0, it indicates that the atlas with the atlas ID J does not have the associated geometry video data. If this value is 1, it indicates having.
  • Attribute video present flag (vps_attribute_video_present_flag[ j ]): If this value is 0, it indicates that the atlas with the atlas ID J does not have the associated attribute video data. If this value is 1, it indicates having.
  • 29 is attribute information according to embodiments.
  • Attribute count (ai_attribute_count[ j ]): Indicates the number of attributes associated with the atlas with atlas ID j. ai_attribute_count[ j ] may range from 0 to 127.
  • Attribute type id indicates the attribute type of the attribute video data unit with index i for the atlas whose atlas ID is j.
  • V3C attribute types V3C attribute types:
  • the identifier is ATTR_TEXTURE and the type is Texture.
  • ID is 1
  • ID is 1
  • ID is 1
  • ID is 1
  • ID is 1
  • ID is 1
  • ID is 1
  • ID is 1
  • ID is 1
  • ID is 1
  • ID is 1
  • ID is 1
  • ID is 1
  • ID is 1
  • ID is 1
  • ID is 1
  • ID is 1
  • ID is 1
  • ID is 1
  • ID is 1
  • ID is 1
  • ID is 1
  • ID is 1
  • type is Material ID.
  • the identifier is ATTR_TRANSPARENCY and the type is Transparency.
  • the ID is 3
  • the identifier is ATTR_REFLECTANCE and the type is Reflectance.
  • the ID is 4, the identifier is ATTR_NORMAL, and the type is Normals.
  • the identifier is ATTR_RESERVED and the type is Reserved.
  • the identifier is ATTR_UNSPECIFIED and the type is Unspecified.
  • ATTR_TEXTURE Indicates an attribute including texture information of a volumetric frame. For example, it may represent a property including RGB (Red, Green, Blue) color information.
  • ATTR_MATERIAL_ID Represents an attribute containing additional information identifying the material type of a point in a volumetric frame.
  • the material type can be used as an indicator to identify an object or property of a point within a volumetric frame.
  • ATTR_TRANSPARENCY Indicates an attribute that includes transparency information associated with each point of the volumetric frame.
  • ATTR_REFLECTANCE Indicates an attribute including reflectance information associated with each point of the volumetric frame.
  • ATTR_NORMAL Indicates an attribute including unit vector information related to each point of the volumetric frame.
  • a unit vector specifies the direction perpendicular to the surface at a point (ie, the direction the point is facing).
  • An attribute frame with this attribute type may have ai_attribute_dimension_minus1 equal to 2.
  • Each channel of an attribute frame with this attribute type may contain one component of the unit vector (x, y, z). Here, the first component contains the x-coordinate, the second component contains the y-coordinate, and the third component contains the z-coordinate.
  • ATTR_UNSPECIFIED It can represent a specific meaningless value. Values marked ATTR_RESERVED are reserved for future use by ISO/IEC and may not exist in bitstreams conforming to this version of this document.
  • Attribute codec ID (ai_attribute_codec_id[ j ][ i ]): Indicates the identifier of the codec used for compressing attribute video data with index i for the atlas whose atlas ID is j. ai_attribute_codec_id[ j ][ i ] may be in the range 0 to 255.
  • Oscillator attribute codec id (ai_auxiliary_attribute_codec_id[ j ][ i ]): RAW and/or EOM coded point of attribute i when the RAW and/or EOM coded point is encoded into an auxiliary video stream to the atlas with atlas ID j Indicates the identifier of the codec used to compress the attribute for video data.
  • ai_auxiliary_attribute_codec_id[ j ][ i ] may be in the range of 0 to 255.
  • ai_auxiliary_attribute_codec_id[ j ][ i ] is inferred to be the same as ai_attribute_codec_id[ j ][ i ].
  • Attribute Map Absolute Coding Presence Flag (ai_attribute_map_absolute_coding_persistence_flag[ j ][ i ]): If this value is 1, it indicates that for the attribute with index i, all attribute maps corresponding to the atlas with atlas ID j are coded without any form of map prediction. indicates. ai_attribute_map_absolute_coding_persistence_flag[ j ][ i ] equal to 0 indicates that the attribute map of the attribute with index i corresponding to the atlas with atlas ID j should use the same map prediction method used for the geometry component of the atlas with the atlas. When ai_attribute_map_absolute_coding_persistence_flag[ j ][ i ] does not exist, its value may be inferred to be equal to 1.
  • a 3D array AttributeMapAbsoluteCodingEnabledFlag indicating whether a particular map of an attribute should be coded with or without prediction is obtained as follows:
  • AttributeMapAbsoluteCodingEnabledFlag[ j ][ i ][ k ] 1
  • AttributeMapAbsoluteCodingEnabledFlag[ j ][ i ][ k ]
  • Attribute dimension (ai_attribute_dimension_minus1[ j ][ i ]): If 1 is added to this value, it indicates the total number of dimensions (ie, number of channels) of the attribute with index i for the atlas with atlas ID j. ai_attribute_dimension_minus1[ j ][ i ] may be in the range of 0 to 63.
  • Attribute dimension partitions (ai_attribute_dimension_partitions_minus1[ j ][ i ]): If 1 is added to this value, the attribute channel of the attribute with index i indicates the number of partition groups that should be grouped for the atlas with atlas ID j. ai_attribute_dimension_partitions_minus1[ j ][ i ] may be in the range 0 to 63.
  • Attribute partition channels (ai_attribute_partition_channels_minus1[ k ][ i ][ j ]): If we add 1 to this value, the number of channels assigned to the dimension partition group with index j of the attribute with index i for the atlas with atlas ID k indicates ai_attribute_partition_channels_minus1[ k ][ i ][ j ] MAY range from 0 to ai_attribute_dimension_minus1[ k ][ i ] for all dimension partition groups.
  • Attribute MSB align flag (ai_attribute_MSB_align_flag[ j ][ i ]): Indicates how the decoded attribute video sample with attribute index i for the atlas with atlas ID j is converted into a sample of attribute bit depth.
  • Attribute count (ai_attribute_count[ j ]): Indicates the number of attributes associated with the atlas whose atlas ID is j. ai_attribute_count[ j ] may be in the range 0 to 127.
  • Attribute SLF camera view selection type (ai_attribute_slf_cameraview_selection_type[ j ]_: Indicates whether the color information of SLF data is compressed and how to select camera viewpoints for each patch used when SLF data is compressed. If this value is 0, it is a general non-SLF data. Indicates that the attribute information of the point cloud data is transmitted If this value is 1 to 3, it indicates that the attribute information obtained from the SLF data is transmitted, and if it is 1, it is a method of selecting a camera viewpoint for each patch when generating an attribute video according to the embodiments.
  • Attribute merged video flag (ai_attribute_merged_video_flag[j][i]): If this value is 1, it indicates that the transmitted Attribute video is a synthesized Attribute video generated using the Texture video generation method 2 according to the embodiments. If this value is 0, it indicates that each of the K attribute videos generated as described in Texture video creation method 1 are transmitted.
  • FIG. 30 shows a video synthesis structure according to embodiments.
  • 30 is an example of a video synthesis structure when six attribute videos are synthesized into one merged attribute video and transmitted.
  • 31-32 show atlas sequence parameter sets according to embodiments.
  • Atlas sequence parameter set id (asps_atlas_sequence_parameter_set_id): Provides an identifier for the atlas sequence parameter set for reference in other syntax elements.
  • Frame width (asps_frame_width): It represents the width of the atlas frame as an integer number of samples.
  • the sample corresponds to the luma sample of the video component.
  • the V3C bitstream conformance requirement that the value of asps_frame_width must be equal to the value of vps_frame_width[ j ] may be followed. where j is the ID of the current map.
  • Frame height represents the atlas frame height as an integer number of samples.
  • the sample corresponds to the luma sample of the video component. It may follow the requirement of V3C bitstream conformance that the value of asps_frame_height must be equal to the value of vps_frame_height[ j ]. where j is the ID of the current map.
  • Geometry 3d bit depth (asps_geometry_3d_bit_depth_minus1): If 1 is added to this value, it indicates the bit depth of the geometric coordinates of the reconstructed volumetric content.
  • asps_geometry_3d_bit_depth_minus1 may range from 0 to 31.
  • Geometry 2d bit depth (asps_geometry_2d_bit_depth_minus1): Add 1 to this value to indicate the bit depth of the geometry when projected onto a 2D image.
  • asps_geometry_2d_bit_depth_minus1 may range from 0 to 31.
  • Attribute used camera view number flag (asps_attribute_use_fixed_number_of_cameraview_flag): When a camera viewpoint is selected for each patch using the method according to the embodiments and an attribute video is generated through this, the number of camera viewpoints selected for all patches in the sequence If are equal, this value can have 1. On the other hand, if the number of camera viewpoints for each selected patch is applied differently in units of frames, tiles, or patches, this value may have 0.
  • the value of asps_log2_max_atlas_frame_order_cnt_lsb_minus4 may be in the range of from 0 to 12 .
  • Adding 1 to this value specifies the maximum required size of the decoded atlas frame buffer for CAS, in atlas frame storage buffer units.
  • the value of asps_max_dec_atlas_frame_buffering_minus1 may be in the range of 0 to 15.
  • Long Term Reference Atlas Frame Flag (asps_long_term_ref_atlas_frames_flag): If this value is 0, it indicates that the long-term reference atlas frame is not used for inter prediction of the atlas frame coded in CAS.
  • asps_long_term_ref_atlas_frames_flag 1 indicates that a long-term reference atlas frame may be used for inter prediction of one or more coded atlas frames in CAS.
  • Reference atlas frame list number (asps_num_ref_atlas_frame_lists_in_asps): Indicates the number of ref_list_struct( rlsIdx ) syntax structures included in the atlas sequence parameter set.
  • the value of asps_num_ref_atlas_frame_lists_in_asps may be in the range of 0 to 64.
  • Orientation flag (asps_use_eight_orientations_flag): 0, indicates that the patch direction index for the patch with index j in the tile with tile ID i, pdu_orientation_index[ i ][ j ] is in the range of 0 to 1 (inclusive).
  • asps_use_eight_orientations_flag 1 indicates that pdu_orientation_index[ i ][ j ] is in the range from 0 to 7.
  • Projection enable flag (asps_extended_projection_enabled_flag): If 0, indicates that patch projection information is not signaled for the current atlas tile. If 1, asps_extended_projection_enabled_flag indicates that patch projection information is signaled for the current atlas tile. If asps_extended_projection_enabled_flag is not present, the value is inferred to be 0.
  • Projection max number (asps_max_number_projections_minus1): If 1 is added to this value, it represents the maximum value that can be expressed for the patch projection ID syntax element pdu_projection_id[ i ][ j ] for the patch with index j in the tile with tile ID i . If asps_max_number_projections_minus1 does not exist, its value is inferred to be equal to 5.
  • Normal axis limit quantization enable flag (asps_normal_axis_limits_quantization_enabled_flag): If 1, indicates that a quantization parameter should be signaled and used to quantize normal axis related elements of a patch data unit, a merged patch data unit, or an inter-patch data unit.
  • asps_normal_axis_limits_quantization_enabled_flag is equal to 0, quantization is not applied to normal axis related elements of patch data units, merged patch data units, or inter-patch data units.
  • Normal axis max delta value enable flag (asps_normal_axis_max_delta_value_enabled_flag): If 1, the maximum normal shift value of the normal axis that can exist in the geometry of the patch with index i in the frame with index j is in each patch data unit and merged patch data. Indicates that it will be displayed in the bitstream for Units or interpatch data units. If asps_normal_axis_max_delta_value_enabled_flag is 0, the maximum normal shift value of the normal axis that can exist in the geometry of the patch with index i in the frame with index j is in the bitstream for each patch data unit, merged patch data unit, or inter-patch data unit. may not be displayed.
  • Patch Precedence Order Flag (asps_patch_precedence_order_flag): Indicates the patch priority used to assign atlas samples to patches. If asps_patch_precedence_order_flag is 1, it indicates that the patch priority for the current atlas is the same as the decoding order of the patch. Asps_patch_precedence_order_flag equal to 0 indicates that the patch priority for the current atlas is the reverse of the decoding order of the patch.
  • asps_log2_patch_packing_block_size can range from 0 to 7.
  • Patch size quantization present flag (asps_patch_size_quantizer_present_flag): If 1, indicates that the patch size quantization parameter is in the atlas tile header. If asps_patch_size_quantizer_present_flag is equal to 0, the patch size quantization parameter may not be present.
  • Map Count (asps_map_count_minus1): Add 1 to this value to indicate the number of maps available to encode the geometry and attribute data of the current atlas.
  • asps_map_count_minus1 may range from 0 to 15.
  • the bitstream conformance requirement for this document is that asps_map_count_minus1 equals vps_map_count_minus1[ atlasID ], where atlasID is the map ID of the current map.
  • Pixel Deinterleaving Enable Flag (asps_pixel_deinterleaving_enabled_flag): equal to 1 indicating that decoded geometry and attribute data may require an additional spatial interpolation process during reconstruction.
  • asps_pixel_deinterleaving_enabled_flag 0 indicates that no additional spatial interpolation process is required for decoded geometry and attribute data.
  • Map pixel deinterleaving flag (asps_map_pixel_deinterleaving_flag[i]): equal to 1 indicates that an additional spatial interpolation process should be performed on the associated geometry and attribute data of the projected patch from the map with index i in the current map.
  • asps_map_pixel_deinterleaving_flag[ i ] equal to 0 indicates that no additional spatial interpolation process is performed. If it does not exist, the value of asps_map_pixel_deinterleaving_flag[ i ] is inferred to be 0.
  • Raw patch enable flag (asps_raw_patch_enabled_flag): If 1, it indicates that information related to RAW coded points is included in the decoded geometry and attribute video for the current atlas. asps_raw_patch_enabled_flag equal to 0 indicates that the decoded geometry and attribute video does not contain information related to RAW coded points.
  • EOM patch enable flag (asps_eom_patch_enabled_flag): If 1, indicates that the decoded occupied video for the current atlas contains information related to whether an intermediate depth position between two depth maps is occupied. asps_eom_patch_enabled_flag equal to 0 indicates that the decoded occupied video does not contain information related to whether an intermediate depth position between two depth maps is occupied. If asps_eom_patch_enabled_flag is equal to 1, it may follow the requirement of bitstream conformance that oi_lossy_occupancy_compression_threshold[ atlasID ] must be equal to 0.
  • EOM fix bit count (asps_eom_fix_bit_count_minus1): If 1 is added to this value, it indicates the bit unit size of the EOM codeword.
  • asps_eom_fix_bit_count_minus1 may be in the range of 0 to Min(15, oi_occupancy_2d_bit_depth_minus1[atlasID ] - 1). where atlasID is the map ID of the current map.
  • Oscillator video enable flag (]asps_auxiliary_video_enabled_flag): If 1, it indicates that information related to RAW and EOM patch types can be placed in the auxiliary video lower bitstream.
  • asps_auxiliary_video_enabled_flag 0 indicates that information related to RAW and EOM patch types can be placed only in the basic video lower bitstream.
  • vps_auxiliary_video_present_flag[atlasID] (where atlasID is the atlas ID of the current atlas) is 0, asps_auxiliary_video_enabled_flag may conform to the requirement of bitstream conformance equal to 0.
  • Point local reconstruction enable flag (asps_plr_enabled_flag): If 1, it indicates that point local reconstruction mode information may be present in the bitstream for the current atlas. asps_plr_enabled_flag equal to 0 indicates that information related to the point local reconstruction mode is not present in the bitstream for the current atlas.
  • asps_plr_enabled_flag be equal to 0 when asps_pixel_deinterleaving_enabled_flag is equal to 1.
  • Parameter presence flag (asps_vui_parameters_present_flag): If 1, it indicates that the vui_parameters() syntax structure exists. asps_vui_parameters_present_flag equal to 0 indicates that the vui_parameters() syntax structure does not exist.
  • Extension present flag (asps_extension_present_flag): If 1, indicates that the syntax elements asps_vpcc_extension_present_flag and asps_extension_7bits are present in the atlas_sequence_parameter_set_rbsp syntax structure. asps_extension_present_flag equal to 0 indicates that the syntax elements asps_vpcc_extension_present_flag and asps_extension_7bits are not present.
  • VPCC extension presence flag (asps_vpcc_extension_present_flag): If 1, indicates that the asps_vpcc_extension() syntax structure is in the atlas_sequence_parameter_set_rbsp syntax structure. asps_vpcc_extension_present_flag equal to 0 indicates that this syntax structure does not exist. When not present, the value of asps_vpcc_extension_present_flag is inferred to be equal to 0.
  • Extension bits (asps_extension_7bits): If 0, indicates that the asps_extension_data_flag syntax element does not exist in the ASPS RBSP syntax structure. When present, asps_extension_7bits may be zero in bitstreams conforming to this version of this document. A non-zero value of asps_extension_7bits may be reserved for future use in ISO/IEC. The decoder shall allow the value of asps_extension_7bits to be a value other than 0, and may ignore all asps_extension_data_flag syntax elements in ASPS NAL units. If it does not exist, the value of asps_extension_7bits is inferred to be equal to 0.
  • Extension data flag (asps_extension_data_flag): It can have any value. Its existence and value do not affect decoder conformance to the profile specified in this version of this document. Decoders conforming to this version of this document MAY ignore all asps_extension_data_flag syntax elements.
  • 33 shows an atlas frame parameter set according to embodiments.
  • Atlas frame parameter set id (apps_atlas_frame_parameter_set_id): Identifies the atlas frame parameter set for reference in other syntax elements.
  • Atlas sequence parameter set ID (apps_atlas_sequence_parameter_set_id): Indicates an asps_atlas_sequence_parameter_set_id value for the active atlas sequence parameter set.
  • Output flag presence flag (afps_output_flag_present_flag): If 1, it indicates that the ath_atlas_output_flag syntax element is present in the associated tile header. afps_output_flag_present_flag equal to 0 indicates that the ath_atlas_output_flag syntax element is not present in the associated tile header.
  • Number of reference indices (apps_num_ref_idx_default_active_minus1): If 1 is added to this value, it indicates an inferred value of the variable NumRefIdxActive for a tile whose ath_num_ref_idx_active_override_flag is 0.
  • the value of afps_num_ref_idx_default_active_minus1 may have a range from 0 to 14.
  • MaxLtAtlasFrmOrderCntLsb 2 * ( +asps_add2_max_atlas_fps_frame_minus_cnt4).
  • the value of afps_additional_lt_afoc_lsb_len may be in the range of 0 to 28 (including asps_log2_max_atlas_frame_order_cnt_lsb_minus4).
  • afps_additional_lt_afoc_lsb_len may be equal to 0.
  • LOD mode enable flag (apps_lod_mode_enabled_flag): 1 indicates that the LOD parameter can be present in the patch.
  • afps_lod_mode_enabled_flag 0 indicates that the LOD parameter should not be present in the patch.
  • Raw3D Offset Bit Count Mode Flag (apps_raw_3d_offset_bit_count_explicit_mode_flag): If 1, rpdu_3d_offset_u[ tileID ][ p ], rpdu_3d_offset_v[ tileID and ] Indicates that bits in the fixed length representation of [p] are explicitly coded by ath_raw_3d_offset_axis_bit_count_minus1 of the atlas tile header referring to afps_atlas_frame_parameter_set_id. afps_raw_3d_offset_bit_count_explicit_mode_flag equal to 0 indicates that the value of ath_raw_3d_offset_axis_bit_count_minus1 is implicitly derived.
  • Extension present flag (afps_extension_present_flag): If 1, indicates that the afps_extension_8bits syntax element is present in the atlas_frame_parameter_set_rbsp syntax structure. afps_extension_present_flag equal to 0 indicates that the syntax element afps_extension_8bits is not present. The value of afps_extension_present_flag may be 0 in this version of this document.
  • Extension bits (afps_extension_8bits): If 0, indicates that the afps_extension_data_flag syntax element does not exist in the AFPS RBSP syntax structure. If present, afps_extension_8bits MAY be zero in bitstreams conforming to this version of this document. A non-zero value of afps_extension_8bits may be reserved for future use in ISO/IEC. The decoder may allow the afps_extension_8bits value as a value other than 0 and ignore all afps_extension_data_flag syntax elements in AFPS NAL units. If it does not exist, the value of afps_extension_8bits is inferred to be equal to 0.
  • Extension data flag may have any value. Its existence and value do not affect decoder conformance to the profile specified in this version of this document. Decoders conforming to this version of this document MAY ignore all afps_extension_data_flag syntax elements.
  • 34-35 show atlas frame tile information according to embodiments.
  • afti_single_tile_in_atlas_frame_flag Single tile flag in atlas frame: If 1, it indicates that there is only one tile in each atlas frame referring to AFPS. afti_single_tile_in_atlas_frame_flag equal to 0 indicates that there may be one or more tiles in each atlas frame referencing AFPS.
  • afti_uniform_partition_spacing_flag Uniform Partition Spacing Flag: If 1, indicates that the tile partitioning of the atlas uses a method of evenly distributing column and row partition boundaries throughout the atlas frame. Information corresponding to these boundaries is signaled using the afti_partition_cols_width_minus1 and afti_partition_rows_height_minus1 syntax elements, respectively. afti_uniform_partition_spacing_flag equal to 0 indicates that the tile partitioning of the atlas uses a method that can create column and row partition boundaries that may or may not be evenly distributed throughout the atlas frame.
  • these boundaries are signaled using the syntax elements afti_num_partition_columns_minus1 and afti_num_partition_rows_minus1 and the list of syntax element pairs afti_partition_column_width_minus1[ i ] and afti_partition_row_height_minus1[ i ].
  • the value of afti_uniform_partition_spacing_flag is inferred to be equal to 1.
  • Partition column width (afti_partition_cols_width_minus1): If 1 is added to this value, when afti_uniform_partition_spacing_flag is 1, it indicates the width of the tile partition column excluding the rightmost tile partition column of the atlas frame. If it does not exist, the value of afti_partition_cols_width_minus1 is inferred to be equal to asps_frame_width / 64-1.
  • Partition row height (afti_partition_rows_height_minus1): If 1 is added to this value, when afti_uniform_partition_spacing_flag is 1, the height of the tile partition row excluding the bottom tile partition row of the atlas frame is expressed in units of 64 samples. inclusive. If it does not exist, the value of afti_partition_rows_height_minus1 is inferred to be equal to asps_frame_height/64-1.
  • afti_num_partition_columns_minus1 Number of partition columns (afti_num_partition_columns_minus1): If 1 is added to this value, it indicates the number of tile partition columns used to divide the atlas frame when afti_uniform_partition_spacing_flag is equal to 0.
  • the value of afti_num_partition_columns_minus1 may be in the range of 0 to asps_frame_width / 64-1. If the specified afti_single_tile_in_atlas_frame_flag is equal to 1, the value of afti_num_partition_columns_minus1 is inferred to be equal to 0.
  • afti_num_partition_rows_minus1 Number of partition rows (afti_num_partition_rows_minus1): If 1 is added to this value, it indicates the number of tile partition rows used to divide the atlas frame when afti_uniform_partition_spacing_flag is equal to 0.
  • the value of afti_num_partition_rows_minus1 may be in the range of 0 to asps_frame_height / 64-1 (inclusive). If afti_single_tile_in_atlas_frame_flag is equal to 1, it is inferred that the value of afti_num_partition_rows_minus1 is equal to 0.
  • NumPartitionsInAtlasFrame variable is set equal to NumPartitionColumns * NumPartitionRows.
  • NumPartitionsInAtlasFrame may be greater than 1.
  • Partition column width (afti_partition_column_width_minus1[ i ]): If 1 is added to this value, the width of the i-th tile partition column is expressed in units of 64 samples.
  • Partition row height (afti_partition_row_height_minus1[ i ]): If 1 is added to this value, the height of the i-th tile partition row is expressed in units of 64 samples.
  • afti_single_partition_per_tile_flag Single partition flag per tile: If 1, it indicates that each tile referring to this AFPS includes one tile partition. afti_single_partition_per_tile_flag equal to 0 indicates that a tile referencing this AFPS may include one or more tile partitions. When not present, the value of afti_single_partition_per_tile_flag is inferred to be equal to 1.
  • afti_num_tiles_in_atlas_frame_minus1 may be in the range of 0 to NumPartitionsInAtlasFrame-1.
  • afti_single_partition_per_tile_flag is equal to 1
  • the value of afti_num_tiles_in_atlas_frame_minus1 is inferred to be equal to NumPartitionsInAtlasFrame-1.
  • Top left partition index indicates the partition index of the tile partition located at the upper left corner of the i-th tile.
  • the value of afti_top_left_partition_idx[ i ] may range from 0 to NumPartitionsInAtlasFrame-1. If it does not exist, the value of afti_top_left_partition_idx[ i ] is inferred to be equal to i.
  • the length of the afti_top_left_partition_idx[i] syntax element is Ceil(Log2(NumPartitionsInAtlasFrame) bits.
  • Bottom right partition column offset (afti_bottom_right_partition_column_offset[ i ]): Indicates the offset between the column position of the tile partition located at the lower right corner of the i-th tile and the column position of the tile partition whose partition index is equal to afti_top_left_partition_idx[ i ].
  • afti_single_partition_per_tile_flag is equal to 1
  • the value of afti_bottom_right_partition_column_offset[i] is inferred to be equal to 0.
  • Bottom right partition row offset (afti_bottom_right_partition_row_offset[ i ]): Indicates the offset between the column position of the tile partition located at the lower right corner of the i-th tile and the column position of the tile partition whose partition index is equal to afti_top_left_partition_idx[ i ].
  • afti_single_partition_per_tile_flag is equal to 1
  • the value of afti_bottom_right_partition_column_offset[i] is inferred to be equal to 0.
  • topLeftColumn[ i ] topLeftRow[ i ]
  • bottomRightColumn[ i ] and bottomRightRow[ i ] specifying the tile column and row positions corresponding to the top-left and bottom-right tiles of the tile can be computed as follows:
  • topLeftColumn[ i ] afti_top_left_partition_idx[ i ] % NumPartitionColumns
  • topLeftRow[ i ] afti_top_left_partition_idx[ i ] / NumPartitionColumns
  • bottomRightColumn[ i ] topLeftColumn[ i ] + afti_bottom_right_partition_column_offset[ i ]
  • bottomRightRow[ i ] topLeftRow[ i ] + afti_bottom_right_partition_row_offset[ i ]
  • bottomRightColumn[ i ] and bottomRightRow[ i ] values must be less than or equal to ( asps_frame_width + 63 ) / 64-1 and ( asps_frame_height + 63 ) / 64-1 respectively.
  • the TileOffsetX[ i ], TileOffsetY[ i ], TileWidth[ i ], and TileHeight[ i ] variables specifying the horizontal position, vertical position, width, and height of the tile, respectively, are calculated as PartitionWidth [ i ] and PartitionHeight[ j ] as can be:
  • TileOffsetX[ i ] PartitionPosX[ topLeftColumn[ i ] ]
  • TileOffsetY[ i ] PartitionPosY[ topLeftColumn[ i ] ]
  • TileWidth[ i ] + PartitionWidth[ j ]
  • Attribute use fixed number of camera views flag (afti_attribute_use_fixed_number_of_cameraview_flag): When a camera viewpoint for each patch is selected using the method according to the embodiments and an attribute video is generated through this, the number of camera viewpoints selected for all patches in the tile is If the number is the same, this value can be 1. On the other hand, if the number of camera viewpoints selected for each patch is applied differently, this value may be 0.
  • Oscillator video tile row width (afti_auxiliary_video_tile_row_width_minus1): If 1 is added to this value, it indicates the nominal width of all auxiliary video sub-bitstreams in units of 64 integer samples. When afti_auxiliary_video_tile_row_width_minus1 does not exist, its value may be inferred to be equal to -1.
  • Oscilry video tile row height (afti_auxiliary_video_tile_row_height[ i ]): Indicates the nominal height in units of 64 integer samples of the i-th vertical sub-region in each auxiliary video sub-bitstream associated with the i-th tile of the atlas. When afti_auxiliary_video_tile_row_height[ i ] does not exist, its value may be inferred to be equal to 0.
  • each sub-region associated with the i-th tile of the atlas AuxTileHeight[ i ] can be calculated as follows:
  • AuxTileHeight[ i ] afti_auxiliary_video_tile_row_height[ i ] * 64
  • AuxTileOffset[ i ] AuxTileOffset[ i - 1 ] + AuxTileHeight[ i - 1 ], for all i > 0.
  • the nominal width AuxVideoWidthNF and height AuxVideoHeightNF of all auxiliary video lower bitstreams associated with the atlas can be calculated as follows:
  • AuxVideoWidthNF ( afti_auxiliary_video_tile_row_width_minus1 + 1 ) * 64
  • N is equal to afti_num_tiles_in_atlas_frame_minus1
  • Signaled tile ID flag (afti_signalled_tile_id_flag): 1 indicates that there is a tile ID for each tile. afti_signalled_tile_id_flag equal to 0 indicates that the tile ID is not signaled.
  • Signaled tile ID length (afti_signalled_tile_id_length_minus1): If 1 is added to this value, it indicates the number of bits used to indicate the syntax element afti_tile_id[ i ] and the syntax element ath_id of the tile header, if present.
  • the value of afti_signalled_tile_id_length_minus1 may be in the range of 0 to 15. If not present, the value of afti_signalled_tile_id_length_minus1 is inferred to be equal to Ceil(Log2(afti_num_tiles_in_atlas_frame_minus1+1)-1.
  • Tile ID (afti_tile_id[ i ]): Indicates the tile ID of the i-th tile.
  • the length of the afti_tile_id[ i ] syntax element is afti_signalled_tile_id_length_minus1+1 bits.
  • the value of afti_tile_id[i] may be inferred to be equal to i for each i in the range from 0 to afti_num_tiles_in_atlas_frame_minus1.
  • the length of the afti_tile_id[ i ] syntax element is afti_signalled_tile_id_length_minus1+1 bits.
  • the variable FirstTileID can be computed as:
  • FirstTileID Min(FirstTileID, afti_tile_id[ i ])
  • TileIDToIndex and TileIndexToID arrays may provide forward and backward mapping of IDs associated with each tile, respectively, and may provide an ordinal index of how each tile is specified in the atlas frame tile information syntax.
  • Atlas adaptation parameter set ID (aaps_atlas_adaptation_parameter_set_id): can identify the atlas adaptation parameter set for reference in other syntax elements.
  • Attribute use fixed number of camera views (aaps_attribute_use_fixed_number_of_camerview_flag): When a camera viewpoint for each patch is selected using the method according to the embodiments and an attribute video is generated through this, the same for all patches in the tile unit referring to the corresponding atlas_adaptation_parameter_set_rbsp syntax. When the number of camera viewpoints is selected and this value is defined in the atlas_adaptation_parameter_set_rbsp syntax, this value is signaled as 1. However, the number is applied differently in units of frames, tiles, or patches, and 0 is signaled when the value is defined in another syntax structure.
  • Log max presence flag (aaps_log2_max_afoc_present_flag): 1 indicates that the syntax element aaps_log2_max_atlas_frame_order_cnt_lsb is present in the atlas_adaptation_parameter_set_rbsp syntax structure.
  • aaps_log2_max_afoc_present_flag 0 indicates that the syntax element aaps_log2_max_atlas_frame_order_cnt_lsb is not present.
  • Log max atlas frame order count (aaps_log2_max_atlas_frame_order_cnt_lsb_minus4): The value of the variable MaxAtlasFrmOrderCntLsb used in the decoding process for the frame order count can be calculated as follows:
  • MaxAtlasFrmOrderCntLsb 2 ( aaps_log2_max_atlas_frame_order_cnt_lsb_minus4 + 4 )
  • aaps_log2_max_atlas_frame_order_cnt_lsb_minus4 may be in the range of 0 to 12.
  • the value of MaxAtlasFrmOrderCntLsb may require bitstream conformance which must be the same for all atlas lower bitstreams of CVS.
  • Extension presence flag (aaps_extension_present_flag): 1 indicates that the aaps_vpcc_extension_present_flag and aaps_extension_7bits syntax elements are present in the atlas_adaptation_parameter_set_rbsp syntax structure.
  • aaps_extension_present_flag 0 indicates that the syntax elements aaps_vpcc_extension_present_flag and aaps_extension_7bits are not present.
  • VPCC extension presence flag (aaps_vpcc_extension_present_flag): 1 indicates that the aaps_vpcc_extension() syntax structure is in the atlas_adaptation_parameter_set_rbsp syntax structure. aaps_vpcc_extension_present_flag equal to 0 indicates that this syntax structure does not exist. When not present, the value of aaps_vpcc_extension_present_flag may be inferred to be equal to 0.
  • Extension bits (aaps_extension_7bits): 0 specifies that the aaps_extension_data_flag syntax element is not present in the AAPS RBSP syntax structure. If present, aaps_extension_7bits MAY be equal to 0 in bitstreams conforming to this version of this document. A non-zero value of aaps_extension_7bits may be reserved for future use in ISO/IEC. The decoder may allow the aaps_extension_7bits value as a value other than 0 and ignore all aaps_extension_data_flag syntax elements in AAPS NAL units. If it does not exist, the value of aaps_extension_7bits may be inferred to be equal to 0.
  • Extension data flag (aaps_extension_data_flag): may have any value. Its existence and value may not affect decoder conformance to the profile specified in this version of this document. Decoders conforming to this version of this document MAY ignore all aaps_extension_data_flag syntax elements.
  • patch information suitable for the patch mode may be transmitted as shown in FIG. 37 .
  • Tile ID is the same as tileID and can be expressed as a multiple of PatchPackingBlockSize.
  • Tile ID is the same as tileID and can be expressed as a multiple of PatchPackingBlockSize.
  • 3D offsetY(pdu_3d_offset_u[ tileID ][ p ]) indicates the shift to be applied to the reconstructed patch point in the patch with the index p of the current atlas tile whose tile ID is equal to tileID along the tangent axis.
  • the value of pdu_3d_offset_u[ tileID ][ p ] may be in the range of 0 to 2 ⁇ (asps_geometry_3d_bit_depth_minus1+1)-1 (inclusive).
  • the number of bits used to represent pdu_3d_offset_u[tileID][p] may be asps_geometry_3d_bit_depth_minus1+1.
  • 3D offsetV(pdu_3d_offset_v[tileID][p]) indicates the shift to be applied to the reconstructed patch point from the patch with the index p of the current atlas tile whose tile ID is equal to tileID along the bidirectional tangential axis.
  • the value of pdu_3d_offset_v[tileID][p] may be in the range of 0 to 2 ⁇ (asps_geometry_3d_bit_depth_minus1+1)-1 (inclusive).
  • the number of bits used to represent pdu_3d_offset_v[tileID][p] is asps_geometry_3d_bit_depth_minus1+1.
  • Tile ID is equal to tileID and along the normal axis
  • Pdu3dOffsetD[ tileID ][ p ] is:
  • Pdu3dOffsetD[ tileID ][ p ] pdu_3d_offset_d[ tileID ][ p ] ⁇ ath_pos_min_d_quantizer
  • Pdu3dOffsetD[ tileID ][ p ] may be in the range of 0 to 2 ⁇ (asps_geometry_3d_bit_depth_minus1+1)-1 (inclusive).
  • the number of bits used to represent pdu_3d_offset_d[tileID][p] may be equal to (asps_geometry_3d_bit_depth_minus1 - ath_pos_min_d_quantizer + 1).
  • variable rangeDBitDepth may be set equal to Min(asps_geometry_2d_bit_depth_minus1, asps_geometry_3d_bit_depth_minus1) + 1.
  • the value of Pdu3dRangeD[ ]-[ tileID ][ pDB may be assumed to be Pdu3dRangeD[ ]-[pDB]. If present, the value of Pdu3dRangeD[ tileID ][ p ] may range from 0 to 2 ⁇ (rangeDBitDepth)-1 (inclusive).
  • the number of bits used to represent pdu_3d_range_d[ tileID ][ p ] may be equal to ( rangeDBitDepth - ath_pos_delta_max_d_quantizer ).
  • pdu_projection_id[ tileID ][ p ] specifies the values of the projection mode and normal index to the projection plane for the patch whose tile ID has the index p of the current atlas tile equal to the tileID.
  • the value of pdu_projection_id[ tileID ][ p ] must be in the range 0 to asps_max_number_projections_minus1.
  • the number of bits used to represent pdu_projection_id[tileID][p] may be Ceil(Log2(asps_max_number_projections_minus1+1)).
  • Orientation index (pdu_orientation_index[ tileID ][ p ]): Specifies the patch orientation index for the patch with index p of the current atlas tile with tile ID equal to tileID, as shown in Table 11, which is used to transform the atlas. It is used to determine the transformation matrices Ro and Rs. The coordinates of the patch relative to the local patch coordinate system, expressed in coordinates (u,v), before transformation into 3D spatial coordinates. The number of bits used to indicate pdu_orientation_index[tileID][p] is ( asps_use_eight_orientations_flag? 3 : 1 ).
  • Attribute selected camera view count (pdu_attribute_selected_cameraview_count [tileID] [p]): it is signaled by asps_attribute_use_fixed_number_of_cameraview_flag, afps_attribute_use_fixed_number_of_cameraview_flag, aps_attribute_use_fixed_number_of_camerview_flag, afti_attribute_use_fixed_number_of_cameraview_flag is all 0s means that the number of the camera viewpoints are selected by the patch differently determined for each patch. In this case, the number of camera viewpoints selected for the p-th patch of the current atlas tile whose tile ID is tileID is signaled through this syntax.
  • Attribute Selected camera view (pdu_attr_selected_cameraview [ j ]
  • signal information of each camera viewpoint according to the number of selected camera viewpoints. This syntax indicates information of the j-th camera viewpoint, and this value is the camera viewpoint. It can be the Index of , or it can be another distinguishable factor.
  • LOD enable flag (pdu_lod_enabled_flag[ tileID ][ p ]): Specifies the patch direction index for the patch with index p of the current atlas tile with the same tile ID as tileID, which is shown in Table 11 used to transform the atlas. Used to determine the transformation matrices Ro and Rs as indicated. The coordinates of the patch relative to the local patch coordinate system, expressed in coordinates (u,v), before transformation into 3D space coordinates. The number of bits used to indicate pdu_orientation_index[tileID][p] is ( asps_use_eight_orientations_flag? 3 : 1 ).
  • LODscaleX (pdu_lod_scale_x_minus1[ tileID ][ p ]): Before adding the tile ID to the patch coordinates TilePatch3dOffsetU[ tileID ][ p ], at the local x-coordinate of the point in the patch with index p of the current atlas tile with tile ID tileID Indicates the LOD scaling factor to be applied. If pdu_lod_scale_x_minus1[tileID][p] does not exist, its value may be inferred to be equal to 0.
  • FIG. 40 shows an apparatus for transmitting V-PCC point cloud data according to embodiments.
  • the apparatus of FIG. 40 is a method/apparatus for transmitting point cloud data according to embodiments, as shown in FIG. 1 transmitting apparatus 10000, point cloud video encoder 10002, FIG. 4 encoding process, FIG. 15 video/image encoder, and FIG. 18 transmitting apparatus , the XR device 1730 of FIG. 20 and the like.
  • Each component of the transmission method/apparatus may correspond to hardware, software, a processor coupled with a memory, and/or a combination thereof.
  • the transmitter of FIG. 40 When the SLF sequence data is input, the transmitter of FIG. 40 generates a patch for the input point cloud through a patch generator as in the operation in V-PCC (400000). With respect to the generated patches, a camera viewpoint containing the most meaningful color information for each patch is selected through a camera viewpoint selection unit for each patch proposed (400001).
  • the viewpoint selection unit 400001 may select a viewpoint by selecting the method 1 for selecting a camera viewpoint for each patch or method 2 for selecting a camera viewpoint for each patch according to embodiments.
  • the geometry image and occult map generated for the input point cloud may be the same as in the V-PCC method.
  • the geometry image generating unit 400003 generates a geometry image from the geometry data of the point cloud data.
  • the patch packing unit 400002 packs each generated patch into a 2D image.
  • a suggested texture video is generated based on the patch-packed result.
  • a single texture video is generated by reflecting the camera viewpoint information selected for each patch, or a plurality of texture videos are generated by the Carrera viewpoint-based texture image generator 400004 for each patch. ) can be created by
  • the number of generated videos is determined according to a case in which the largest number of camera viewpoints is selected.
  • the texture image generation unit 400004 determines whether a single texture video or multiple texture videos are to be generated, and the device/user selects a texture video generation method 1 and a texture video generation method 2 according to the embodiments. You can create videos.
  • the generated texture video is compressed and transmitted using a 2D video codec along with the geometry video and occupanci map respectively. Selection methods used when generating a texture video are signaled as side information and transmitted together.
  • the encoding preprocessor may receive a geometry image and may perform preprocessing necessary for encoding by receiving multiple or single texture images (attributes).
  • the video encoder may encode geometry data and/or attribute data.
  • the subcomplemented geometry data may be reconstructed by the geometry restoration unit and used for attribute encoding.
  • the smoother may apply processing such as filtering to the reconstructed geometry data and transmit it to the texture image generator.
  • the metadata enrichment may encode additional patch information related to geometry data (geometric image), attribute data (texture image), and occupancy map.
  • the additional patch information may be generated by the viewpoint selector 400001 .
  • the multiplexer may multiplex the encoded geometry data, attribute data, accuracy map, and additional patch information into a bitstream.
  • the transmitter may transmit the encoded point cloud data.
  • the transmitting device components in FIG. 40 may correspond to the transmitting device components in FIG. 18 and the like.
  • the viewpoint selector 400001 may be connected to the patch generator or included in the patch generator.
  • the texture image generator 400004 may be included in or connected to the patch packing unit 400002 and/or the viewpoint selector 400001 , and may be connected to an encoder of the transmitting apparatus as a packing related preprocessor (processor).
  • processor packing related preprocessor
  • the point cloud data receiving apparatus may restore the point cloud data based on a reverse process of the transmitting apparatus.
  • Receiver (Fig. 2 receiving device 10005, point cloud video decoder 10008, Fig. 16 decoding process, Fig. 17 video/image decoder, Fig. 19 receiving device, Fig. 20 XR device 1730, etc.
  • the component corresponds to hardware, software, a processor connected to the memory, and/or a combination thereof), the input geometry video, texture (attribute) video, occupancy map video and side information data are each decoded to restore the information and , to restore the point cloud of the SLF sequence.
  • the color information of the camera viewpoint representing each point is restored, and through this, the color information of the SLF sequence can be reconstructed or only color information suitable for the purpose can be selected and utilized.
  • Camera viewpoint information corresponding to each patch for each video can be known through the restored additional information, and the restored color information can be applied according to the camera viewpoint condition currently required during point cloud rendering using this. If the currently required camera viewpoint and the camera viewpoint of the target point do not match, color information of the retained camera viewpoint located closest to the required camera viewpoint may be used.
  • the existing method [2] for compressing SLF data using V-PCC causes problems such as requiring multiple video codec instances and high memory usage to compress multiple attribute video data generated during the encoding process. do.
  • the other method [3] proposed to improve this problem excessive data reduction resulted in a side effect of significantly lowering image quality performance.
  • the compression/decompression method has the effect of improving the problem of the SLF sequence compression method and obtaining more efficient compression performance.
  • For a point cloud group divided by patches only meaningful camera viewpoints are selected, and only color information obtained from the corresponding camera viewpoints is transmitted, thereby improving the efficiency of image quality and compression performance.
  • By proposing two methods for selecting meaningful camera viewpoints in units of patches more suitable color information can be transmitted, restored, and utilized according to the purpose of use of the user.
  • the selected camera viewpoint information and transmitted attribute video information are signaled so that the transmitting end and the receiving end can check information on the method. Accordingly, the receiving end can selectively restore only the attribute video required according to the purpose of the user or use color information suitable for the direction to be rendered based on all the restored color information.
  • Operations according to the embodiments described in this document may be performed by a transceiver including a memory and/or a processor according to the embodiments.
  • the memory may store programs (such as a flow chart) for processing/controlling operations according to the embodiments, and the processor may control various operations described in this document.
  • the processor may be referred to as a controller or the like.
  • operations may be performed by firmware, software, and/or a combination thereof, and the firmware, software, and/or a combination thereof may be stored in a processor or stored in a memory.
  • the method/device according to the embodiments uses V-PCC to compress a Surface Light Field (SLF) data set having more attribute properties than a general point cloud data set, a valid camera view for each object included in the SLF data set By transmitting point information, it is intended to improve the point cloud restoration performance for each object at the receiving end.
  • SLF Surface Light Field
  • a method for improving point cloud restoration performance for each object and enabling selective camera viewpoint utilization at the receiving end is provided. This is related to a method of signaling a meaningful camera viewpoint for each object among the camera viewpoints constituting the SLF sequence data in the V-PCC with an SEI message.
  • the method/apparatus according to the embodiments is intended to solve the following problems. Assuming that there is one object in the SLF sequence, we want to remove the restriction that mainly focuses on the compression method.
  • a case in which multiple objects exist in the SLF sequence is considered.
  • an object may be obscured by another adjacent object, resulting in a camera viewpoint in which attribute information cannot be obtained.
  • a camera viewpoint that has not obtained attribute information for an object predicts a value using attribute information obtained from other camera viewpoints, and transmits this value as its own attribute information.
  • the predicted and processed attribute information in other expressions is used to restore the attribute value of another viewpoint at the receiving end, this may affect the deterioration of the image quality of the restored point cloud. Therefore, in the present invention, it is possible to provide an effect of preventing unnecessary restoration of image quality by excluding processed information and restoring attributes using only accurate information.
  • the method/apparatus codes an SLF sequence including multiple objects using V-PCC, a method for transmitting usable camera viewpoint information for each object and a related signaling method are provided.
  • 41 shows an example of an SLF data set including multiple objects according to embodiments.
  • the method/apparatus according to the embodiments may compress and restore point cloud data regarding multiple objects, such as object #1 and object #2.
  • the SLF data set includes all attribute information for each point obtained from a plurality of camera viewpoints as data, and the V-PCC compresses and transmits the SLF data set thus generated. If the SLF data set includes a plurality of objects, as shown in the example of FIG. 41 , a camera viewpoint may occur that is obscured by other adjacent objects and thus cannot acquire attribute information.
  • the right object object #2 in Fig. 41 is obscured by the left object object #1, resulting in two camera viewpoints from which information cannot be obtained (41000), and this is included in the area marked as the occluded area of object #2 It becomes impossible to acquire attribute information through these two camera viewpoints.
  • the attribute information that has not been obtained in this way will be created and used by replacing it with a processed value using the attribute information obtained from other available camera viewpoints.
  • the generated data set is transmitted through V-PCC and used for point cloud restoration.
  • the processed attribute information can replace the information that has not been obtained, but if the attribute values of other surrounding points are restored using this processed information, it may be a factor that reduces the accuracy of the data, and thus the restoration quality may affect the deterioration of
  • the present invention proposes a method of additionally signaling and transmitting information on camera viewpoints usable for each object so that only unprocessed and directly acquired data can be utilized when restoring point attribute information in order to minimize such unnecessary degradation of image quality. do.
  • a method of transmitting camera viewpoint information usable for each object in an SEI message for all objects included in the SLF sequence, we propose a method of transmitting camera viewpoint information usable for each object in an SEI message.
  • Information included in the SEI message is the number of valid camera viewpoints for each object and index information for identifying each camera viewpoint.
  • the hidden region related information according to the embodiments is signaled by being included in the Scene object information SEI message among the Volumetric annotation SEI message family of V-PCC (hereinafter, FIGS. 42-47).
  • An apparatus for transmitting point cloud data may generate information as shown in FIGS. 42-47, include it in a bitstream, and transmit it to a receiving apparatus.
  • An apparatus for receiving point cloud data receives a bitstream including point cloud data, and based on the information of FIGS. 42-47 included in the bitstream, point cloud data (geometric data, attribute data) included in the bitstream , accuracy maps, etc.) can be decoded.
  • This SEI message defines a set of objects that may exist in the volumetric scene and optionally assigns other properties to these objects. These objects can then potentially be associated with other types of information, including patches and 2D volumetric rectangles, which can be defined using patch information and volumetric rectangle information SEI messages.
  • ObjectTracked[ k ] 0 at the beginning of each sequence. where k corresponds to the object index and ranges from 0 to 232 -1. If ObjectTracked[ k ] is 0, indicates that all relevant parameters including object labels, 3D bounding box parameters, priority information, hidden flags, dependency information, visible cone, collision shape, point style, and material ID have default values. indicates.
  • the application can further specify an object index limit.
  • Persistence flag Indicates the persistence of the scene object information SEI message for the current layer. soi_persistence_flag equal to 0 indicates that the scene object information SEI message is applied only to the currently decoded atlas frame.
  • soi_persistence_flag 1 may indicate that the scene object information SEI message persists for the current layer in output order until one of the following conditions is true:
  • Atlas frame aFrmB in the current layer of the coded atlas access unit applicable to the current layer and containing the scene object information SEI message having the same value of Soi_persistence_flag is output, where AtlasFrmOrderCnt( aFrmB ) is greater than AtlasFrmOrderCnt( aFrmA ) Big.
  • AtlasFrmOrderCnt( aFrmB ) and AtlasFrmOrderCnt( aFrmA ) are the AtlasFrmOrderCntVal values of aFrmB and aFrmA immediately after the decoding process call to the atlas frame order count for aFrmB.
  • Reset flag Indicates that information corresponding to this scene object information SEI message is reset to a default value.
  • Object number update Indicates the number of objects to be updated by the current SEI.
  • the value of soi_num_object_updates may range from 0 to 232-1.
  • the default value of soi_num_object_updates is 0.
  • Simple object flag (soi_simple_objects_flag): If equal to 1, it indicates that no additional information about an updated or newly introduced object is signaled. soi_simple_objects_flag equal to 0 indicates that additional information about an updated or newly introduced object can be signaled.
  • Object label presence flag (soi_object_label_present_flag): equal to 1 indicates that object label information is present in the current scene object information SEI message. soi_object_label_present_flag equal to 0 indicates that object label information does not exist.
  • Priority presence flag indicates that priority information is present in the current scene object information SEI message. soi_priority_present_flag equal to 0 indicates that priority information does not exist.
  • Object hidden presence flag (soi_object_hidden_present_flag): equal to 1 indicates that hidden object information is present in the current scene object information SEI message. soi_object_hidden_present_flag equal to 0 indicates that hidden object information does not exist.
  • Visibility cone presence flag (soi_visibility_cones_present_flag): equal to 1 indicates that visibility cone information is present in the current scene object information SEI message. soi_visibility_cones_present_flag equal to 0 indicates that visibility cone information does not exist.
  • 3D bounding box presence flag (soi_3d_bounding_box_present_flag): equal to 1 indicates that 3D bounding box information is present in the current scene object information SEI message. soi_3d_bounding_box_present_flag equal to 0 indicates that 3D bounding box information does not exist.
  • Collision shape presence flag (soi_collision_shape_present_flag): equal to 1 indicates that collision information is present in the current scene object information SEI message. soi_collision_shape_present_flag equal to 0 indicates that collision shape information does not exist.
  • Point style presence flag (soi_point_style_present_flag): equal to 1 indicates that point style information is present in the current scene object information SEI message. soi_point_style_present_flag equal to 0 indicates that point style information does not exist.
  • Material ID presence flag (soi_material_id_present_flag): equal to 1 indicates that material ID information is present in the current scene object information SEI message. soi_material_id_present_flag equal to 0 indicates that material ID information does not exist.
  • Extension presence flag indicates that additional extension information must be present in the current scene object information SEI message. soi_extension_present_flag equal to 0 indicates that additional extension information does not exist. It is a requirement of bitstream conformance for this version of this document that soi_extension_present_flag shall be equal to 0.
  • Object camera view presence flag (soi_object_cameraview_present_flag): If this value is 1, it means that camera viewpoint information for each object is present in the current scene object information SEI message. If this value is 0, the corresponding information does not exist.
  • 3D bounding box scale log (soi_3d_bounding_box_scale_log2): Indicates the scale to be applied to the 3D bounding box parameter that can be specified for the object.
  • 3D bounding box precision (soi_3d_bounding_box_precision_minus8): Add 8 to this value to indicate the precision of the 3D bounding box parameter that can be specified for the object.
  • Logmax object index update (soi_log2_max_object_idx_updated_minus1): If 1 is added to this value, it indicates the number of bits used to signal the object index value in the current scene object information SEI message.
  • Logmax object dependency index (soi_log2_max_object_dependency_idx): Indicates the number of bits used to signal the dependency object index value in the current scene object information SEI message. The default value of soi_log2_max_object_dependency_idx is 0.
  • Object index (soi_object_idx[ i ]): Indicates the object index of the i-th object to be updated.
  • the number of bits used to represent soi_object_idx[ i ] is equal to soi_log2_max_object_idx_updated_minus1 + 1. If there is no soi_object_idx[ i ] in the bitstream, its value is inferred to be equal to 0.
  • Object label update flag (soi_object_label_update_flag[ i ]): If equal to 1, it indicates that there is object label update information for the object having the object index i. soi_object_label_update_flag[ i ] equal to 0 indicates that object label update information does not exist.
  • Object label index (soi_object_label_idx[ i ]): Indicates the label index of the object with index i.
  • the value of soi_object_label_idx[ i ] may be in the range of 0 to 232-1.
  • Priority update flag (soi_priority_update_flag[ i ]): If equal to 1, it indicates that there is priority update information for the object having the object index i. soi_priority_update_flag[ i ] equal to 0 indicates that object priority information does not exist.
  • Priority value Indicates the priority of the object with index i. The lower the priority value, the higher the priority. The default value of soi_priority_value[ i ] is 0.
  • Object hidden flag (soi_object_hidden_flag[ i ]): If equal to 1, it indicates that the object with index i is hidden. soi_object_hidden_flag[ i ] equal to 0 indicates that the object with index i will exist.
  • Object dependency update flag (soi_object_dependency_update_flag[ i ]): If equal to 1, it indicates that there is object dependency update information for the object having the object index i. soi_object_dependency_update_flag[ i ] equal to 0 indicates that object dependency update information does not exist.
  • Object dependency index (soi_object_dependency_idx[ i ][ j ]): Indicates the index of the object having the object index i and the j-th object having the dependency.
  • Visibility cone update flag (soi_visibility_cones_update_flag[ i ]): If equal to 1, it indicates that visibility cone update information exists for the object with object index i. soi_visibility_cones_update_flag[ i ] equal to 0 indicates that visibility cone update information does not exist.
  • DirectionX(soi_direction_x[ i ]) Represents the normalized x-component value of the direction vector for the visibility cone of the object whose object index is i.
  • the value of soi_direction_x[ i ] is inferred to be equal to 1.0 when not present.
  • the default value of soi_direction_x[ i ] is equal to 1.0.
  • DirectionY(soi_direction_y[ i ]) Represents the normalized y-component value of the direction vector for the visibility cone of the object whose object index is i.
  • the value of soi_direction_y[ i ] is inferred to be equal to 1.0 when not present.
  • the default value of soi_direction_y[ i ] is equal to 1.0.
  • Direction Z(soi_direction_z[ i ]) Represents the normalized z-component value of the direction vector for the visibility cone of the object whose object index is i.
  • the value of soi_direction_z[ i ] is inferred to be equal to 1.0 when not present.
  • the default value of soi_direction_z[ i ] is equal to 1.0.
  • Angle(soi_angle[ i ]) The angle of the cone of sight along the direction vector in degrees.
  • the value of soi_angle[ i ] is inferred to be equal to 180 when not present.
  • the default value of soi_angle[ i ] is equal to 180.
  • 3D bounding box update flag (soi_3d_bounding_box_update_flag[ i ]): If equal to 1, it indicates that there is 3D bounding box information for the object with object index i. soi_3d_bounding_box_update_flag[ i ] equal to 0 indicates that 3D bounding box information does not exist.
  • the default value of soi_3d_bounding_box_x[ i ] is 0.
  • the default value of soi_3d_bounding_box_y[ i ] is 0.
  • the default value of soi_3d_bounding_box_z[ i ] is 0.
  • 3D bounding box deltaX ( soi_3d_bounding_box_delta_x[ i ]): Represents the size of the bounding box on the x-axis of the object with index i. The default value of soi_3d_bounding_box_delta_x[ i ] is 0.
  • the default value of soi_3d_bounding_box_delta_y[ i ] is 0.
  • the default value of soi_3d_bounding_box_delta_z[ i ] is 0.
  • Collision shape update flag (soi_collision_shape_update_flag[ i ]): If equal to 1, it indicates that there is collision shape update information for the object with object index i. soi_collision_shape_update_flag[ i ] equal to 0 indicates that no collision shape update information exists.
  • Collision shape ID (soi_collision_shape_id[ i ]): Indicates the collision shape ID of the object with index i. Collision shape IDs are identified through means outside of this document. The default value of soi_collision_shape_id[ i ] is 0.
  • Point style update flag (soi_point_style_update_flag[ i ]): If equal to 1, it indicates that there is point style update information for the object having the object index i. soi_point_style_update_flag[ i ] equal to 0 indicates that point style update information does not exist.
  • Point shape ID (soi_point_shape_id[ i ]): Indicates the point shape ID of the object with index i.
  • the default value of soi_point_shape_id[ i ] is equal to 0.
  • the value of soi_point_shape_id[ i ] MAY be in the range 0 to 2, including bitstreams conforming to this version of this document.
  • Other values of soi_point_shape_id[ i ] may be reserved for future use in ISO/IEC. Decoders conforming to this version of this document may ignore the reserved value of soi_point_shape_id[ i ].
  • Point size (soi_point_size[ i ]): Indicates the point size of the object with index i.
  • the default value of soi_point_size[ i ] is equal to 1.
  • Material ID update flag (soi_material_id_update_flag[ i ]): If equal to 1, it indicates that there is material ID update information for the object with object index i. soi_point_style_update_flag[ i ] equal to 0 indicates that point style update information does not exist.
  • Material ID (soi_material_id[ i ]): Indicates the material ID of the object with index i. The default value of soi_material_id[ i ] is 0. Material IDs are identified through means outside of this document.
  • Object camera view update flag (soi_object_cameraview_update_flag[ i ]): If this value is 1, it indicates that camera viewpoint update information for the i-th object exists. If this value is 0, it indicates that the corresponding information does not exist.
  • Object camera view index (soi_object_cameraview_idx[ i ][ j ]): Indicates the index of the j-th camera viewpoint among the valid camera viewpoints for the ith object. Alternatively, in addition to the index, it can be used by replacing the value of other information that can distinguish the camera viewpoint.
  • a method/apparatus for transmitting point cloud data may encode and transmit point cloud data as follows.
  • the object determines that it is possible to acquire attributes through all camera viewpoints and suggests corresponding camera viewpoint information. It is transmitted using SEI message. However, if an area obscured by a surrounding object occurs and it is impossible to obtain attribute information from a camera viewpoint existing in the corresponding direction (refer to FIG. 41), after excluding them, only information of the remaining available camera viewpoints is sent to the SEI message of the object. to (refer to Figs. 42-47).
  • the obscuration of the camera viewpoint is determined by considering only the situation in which the camera viewpoint is obscured by another object, not the situation in which the camera viewpoint is blocked due to a point area existing in the same object.
  • a method/apparatus for transmitting point cloud data may receive and decode point cloud data as follows.
  • the method checks information of camera viewpoints that can be used for each object and restores attribute data based on them .
  • processed information is excluded when data is restored, unnecessary image quality deterioration can be prevented.
  • a camera viewpoint for each object it is possible to selectively utilize a camera viewpoint for each object according to a user's intention based on information included in the received bitstream. For example, when the user wants to restore only a specific direction of the target object, the camera viewpoints existing within a certain area including the direction are determined, and the information on the actually usable camera viewpoints is checked through the received SEI information. , can be used for rendering. Alternatively, after the user checks valid camera viewpoints from the received information, only specific camera viewpoints determined to be more meaningful among them may be selected and utilized for data restoration.
  • the method/apparatus provides a new camera viewpoint information processing method for an SLF point cloud data set including multiple objects, which was not considered in the SLF sequence compression technique.
  • a camera viewpoint may occur in which attribute information cannot be obtained from each object due to interference between different objects.
  • the embodiments may signal and transmit information on usable camera viewpoints for each object so that only attribute information obtained from actual camera viewpoints can be used when the decoder restores attributes.
  • the reception method/device may prevent unnecessary image quality deterioration by restoring the point cloud by using only attribute information of valid camera viewpoints for each object.
  • it provides the effect of improving the accuracy and flexibility of using attribute information at the receiving end by allowing the necessary camera viewpoints for each object to be selectively used according to the user's intention.
  • FIG. 48 illustrates a method of transmitting point cloud data according to embodiments.
  • the method for transmitting point cloud data may include encoding the point cloud data.
  • the encoding operation may be performed by the transmission device 10000 of FIG. 1, the point cloud video encoder 10002, the file/segment encapsulator 10003, the encoding processor of FIG. 4, the encoder of FIG. 15, the transmission device of FIG. 18, and FIG. 20.
  • the method for transmitting point cloud data may further include transmitting a bitstream including the point cloud data.
  • Transmission operations according to the embodiments include the transmission apparatus 10000 of FIG. 1 , the transmitter 10004 , the transmission of the bitstream of FIG. 4 , the transmission of the bitstream of FIG. 15 , the transmission of the bitstream of the transmission apparatus of FIG. 18 , and the transmission of the XR device 2030 of FIG. 20 .
  • 49 illustrates a method for receiving point cloud data according to embodiments.
  • the method for receiving point cloud data may include receiving a bitstream including point cloud data.
  • Receiving operation is shown in Fig. 1 receiving apparatus 10005, receiver 10006, file/segment decapsulator 10007, Fig. 16-17, bitstream reception including 19 point cloud data, Fig. 20 XR Device 2030: Receiving point cloud data, receiving point cloud data including the data set of FIGS. 21 to 25, receiving data included in the bitstream of FIGS. 26 to 39, receiving point cloud data for multiple objects, FIG. 42 to 47 may include operations such as receiving data included in the bitstream.
  • the method for receiving point cloud data may further include decoding the point cloud data.
  • the decoding operation includes the point cloud video decoder 10008 of FIG. 1 , the decoding of FIGS. 16-17 , the receiving apparatus of FIG. 19 , the decoding of the XR device 2030 of FIG. 20 , and the points including the data sets of FIGS. It may include operations such as cloud data decoding, data decoding included in the bitstreams of FIGS. 26 to 39, point cloud data decoding for multiple objects in FIG. 41, data decoding included in the bitstreams of FIGS. 42 to 47, and the like.
  • a method of transmitting point cloud data may include encoding point cloud data; and transmitting a bitstream including the point cloud data; may include.
  • Point cloud data includes geometric data (position value of a point of an object) obtained from cameras for viewpoints and at least two attributes (eg, a plurality of color values),
  • the method for transmitting point cloud data may further include selecting a specific number of viewpoints from the viewpoints based on an object and a distance between the viewpoints.
  • the method for transmitting point cloud data may further include generating representative attribute information from the selected specific number of viewpoints, and selecting the specific number of viewpoints based on the representative attribute information. This is because it is efficient to first compress the attribute that has a large difference from the representative attribute information (eg, the representative color value of a specific point).
  • texture data may be generated from attribute information for a specific number of viewpoints selected.
  • the texture data may mean a texture (attribute) video or a texture (attribute) image.
  • the method for transmitting point cloud data may further include generating texture data from a specific number of viewpoints based on a difference between the representative attribute information and attribute information for a specific number of viewpoints.
  • the point cloud data transmission method further comprises generating one texture data by merging texture data including attribute information for the selected specific number of viewpoints.
  • a bitstream generated and encoded according to embodiments may include parameter information related to camera view point selection.
  • the point cloud data includes geometric data obtained from objects and at least two attributes, and the bitstream includes valid camera viewpoint-related parameter information generated based on an area occluded between the first object and the second object. may include Since the obscured area may be unnecessary when restoring point cloud data, the bitstream size is reduced by excluding the obscured area, thereby enabling accurate and efficient compression and restoration.
  • An apparatus for receiving point cloud data includes: a receiver for receiving a bitstream including point cloud data; and a decoder for decoding the point cloud data; may include.
  • the point cloud data according to the embodiments includes at least two attributes and geometric data obtained from cameras for viewpoints, and the decoder according to the embodiments provides a view based on an object and a distance between the viewpoints. It is possible to decode attribute data for a specific number of viewpoints selected from ints.
  • the decoder may decode attribute data for a specific number of viewpoints based on representative attribute information generated from the selected specific number of viewpoints.
  • the decoder may decode the texture data generated from attribute information on the selected specific number of viewpoints based on the distance order.
  • the decoder may decode the texture data generated from the specific number of viewpoints based on a difference between the representative attribute information and the attribute information for the specific number of viewpoints.
  • the decoder may decode one piece of texture data in which attribute information for the selected specific number of viewpoints is merged.
  • a bitstream according to embodiments may include parameter information related to camera view point selection.
  • the point cloud data according to the embodiments includes geometric data obtained from objects and at least two attributes, and the bitstream according to the embodiments is a valid camera generated based on an area occluded between the first object and the second object It may include viewpoint-related parameter information.
  • the method/apparatus according to the embodiments enables efficient data compression and restoration without the need to compress and transmit all the plurality of camera information for obtaining the SLF data set in order to provide an accurate representation. Since only a specific number of SLF data with high similarity among a plurality of pieces is efficiently selected and compressed, it is accurate and the compression performance is increased. In addition, since data is compressed by excluding the hidden area, accurate data restoration is possible.
  • Various components of the apparatus of the embodiments may be implemented by hardware, software, firmware, or a combination thereof.
  • Various components of the embodiments may be implemented with one chip, for example, one hardware circuit.
  • the components according to the embodiments may be implemented with separate chips.
  • at least one or more of the components of the device according to the embodiments may be composed of one or more processors capable of executing one or more programs, and the one or more programs may be implemented Any one or more of the operations/methods according to the examples may be performed or may include instructions for performing the operations/methods.
  • Executable instructions for performing the method/acts of the apparatus according to the embodiments may be stored in non-transitory CRM or other computer program products configured for execution by one or more processors, or one or more may be stored in temporary CRM or other computer program products configured for execution by processors.
  • the memory according to the embodiments may be used as a concept including not only volatile memory (eg, RAM, etc.) but also non-volatile memory, flash memory, PROM, and the like. Also, it may be implemented in the form of a carrier wave, such as transmission through the Internet.
  • the processor-readable recording medium is distributed in a computer system connected to a network, so that the processor-readable code can be stored and executed in a distributed manner.
  • first, second, etc. may be used to describe various components of the embodiments. However, interpretation of various components according to the embodiments should not be limited by the above terms. These terms are only used to distinguish one component from another. it is only For example, the first user input signal may be referred to as a second user input signal. Similarly, the second user input signal may be referred to as a first user input signal. Use of these terms should be interpreted as not departing from the scope of the various embodiments. Although both the first user input signal and the second user input signal are user input signals, they do not mean the same user input signals unless the context clearly indicates otherwise.
  • the operations according to the embodiments described in this document may be performed by a transceiver including a memory and/or a processor according to the embodiments.
  • the memory may store programs for processing/controlling operations according to the embodiments, and the processor may control various operations described in this document.
  • the processor may be referred to as a controller or the like.
  • operations may be performed by firmware, software, and/or a combination thereof, and the firmware, software, and/or a combination thereof may be stored in a processor or stored in a memory.
  • the embodiments may be applied in whole or in part to a point cloud data transmission/reception device and system.
  • Embodiments may include modifications/modifications, which do not depart from the scope of the claims and the like.

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Graphics (AREA)
  • Geometry (AREA)
  • Software Systems (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Compression Or Coding Systems Of Tv Signals (AREA)

Abstract

실시예들에 따른 포인트 클라우드 데이터 송신 방법은 포인트 클라우드 데이터를 인코딩하는 단계; 및 포인트 클라우드 데이터를 전송하는 단계; 를 포함할 수 있다. 실시예들에 따른 포인트 클라우드 데이터 수신 방법은 포인트 클라우드 데이터를 수신하는 단계; 포인트 클라우드 데이터를 디코딩하는 단계; 및 포인트 클라우드 데이터를 렌더링하는 단계; 를 포함할 수 있다.

Description

포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법
실시예들은 사용자에게 VR (Virtual Reality, 가상현실), AR (Augmented Reality, 증강현실), MR (Mixed Reality, 혼합현실), 및 자율 주행 서비스 등의 다양한 서비스를 제공하기 위하여 Point Cloud 콘텐츠를 제공하는 방안을 제공한다.
포인트 클라우드는 3D공간 상의 포인트들의 집합이다. 3D공간 상의 포인트들의 양이 많아서 포인트 클라우드 데이터를 생성하기 어려운 문제점이 있다.
포인트 클라우드의 데이터를 전송하고 수신하기 위해서 많은 처리량이 요구되는 문제점이 있다.
실시예들에 따른 기술적 과제는, 전술한 문제점 등을 해결하기 위해서, 포인트 클라우드를 효율적으로 송수신하기 위한 포인트 클라우드 데이터 전송 장치, 전송 방법, 포인트 클라우드 데이터 수신 장치 및 수신 방법을 제공하는데 있다.
실시예들에 따른 기술적 과제는, 지연시간(latency) 및 인코딩/디코딩 복잡도를 해결하기 위한 포인트 클라우드 데이터 전송 장치, 전송 방법, 포인트 클라우드 데이터 수신 장치 및 수신 방법을 제공하는데 있다.
다만, 전술한 기술적 과제만으로 제한되는 것은 아니고, 본 문서 전체 내용에 기초하여 당업자가 유추할 수 있는 다른 기술적 과제로 실시예들의 권리범위가 확장될 수 있다.
상술한 목적 및 다른 이점을 달성하기 위해서 실시예들에 따른 포인트 클라우드 데이터 송신 방법은 포인트 클라우드 데이터를 인코딩하는 단계; 및 포인트 클라우드 데이터를 전송하는 단계; 를 포함할 수 있다.
실시예들에 따른 포인트 클라우드 데이터 수신 방법은 포인트 클라우드 데이터를 수신하는 단계; 포인트 클라우드 데이터를 디코딩하는 단계; 및 포인트 클라우드 데이터를 렌더링하는 단계; 를 포함할 수 있다.
실시예들에 따른 포인트 클라우드 데이터 송신 방법, 송신 장치, 포인트 클라우드 데이터 수신 방법, 수신 장치는 퀄리티 있는 포인트 클라우드 서비스를 제공할 수 있다.
실시예들에 따른 포인트 클라우드 데이터 송신 방법, 송신 장치, 포인트 클라우드 데이터 수신 방법, 수신 장치는 다양한 비디오 코덱 방식을 달성할 수 있다.
실시예들에 따른 포인트 클라우드 데이터 송신 방법, 송신 장치, 포인트 클라우드 데이터 수신 방법, 수신 장치는 자율주행 서비스 등 범용적인 포인트 클라우드 콘텐츠를 제공할 수 있다.
도면은 실시예들을 더욱 이해하기 위해서 포함되며, 도면은 실시예들에 관련된 설명과 함께 실시예들을 나타낸다.
도1은 실시예들에 따른 Point Cloud 콘텐츠 제공을 위한 송신/수신 시스템의 구조의 예시를 나타낸다.
도2는 실시예들에 따른 포인트 클라우드 데이터 갭쳐의 예시를 나타낸다.
도3은 실시예들에 따른 포인트 클라우드 및 지오메트리, 텍스쳐 이미지의 예시를 나타낸다.
도4는 실시예들에 따른 V-PCC 인코딩 처리의 예시를 나타낸다.
도5는 실시예들에 따른 서페이스(Surface)의 탄젠트 플렌(tangent plane) 및 노멀 벡터(normal vector)의 예시를 나타낸다.
도6은 실시예들에 따른 포인트 클라우드의 바운딩 박스(bounding box)의 예시를 나타낸다.
도7은 실시예들에 따른 어큐판시 맵(occupancy map)의 개별 패치(patch) 위치 결정의 예시를 나타낸다.
도8은 실시예들에 따른 노멀(normal), 탄젠트(tangent), 바이탄젠트(bitangent) 축의 관계의 예시를 나타낸다.
도9는 실시예들에 따른 프로젝션 모드의 최소 모드 및 최대 모드의 구성의 예시를 나타낸다.
도10은 실시예들에 따른 EDD 코드의 예시를 나타낸다.
도11은 실시예들에 따른 인접점들의 컬러(color) 값들을 이용한 리컬러링(recoloring)의 예시를 나타낸다.
도12는 실시예들에 따른 푸쉬-풀 백그라운드 필링(push-pull background filling)의 예시를 나타낸다.
도13은 실시예들에 따른 4*4 크기의 블록(block)에 대해 가능한 트라버설 오더(traversal order)의 예시를 나타낸다.
도14는 실시예들에 따른 베스트 트라버설 오더의 예시를 나타낸다.
도15는 실시예들에 따른 2D 비디오/이미지 인코더(2D video/image Encoder)의 예시를 나타낸다.
도16은 실시예들에 따른 V-PCC 디코딩 프로세스(decoding process)의 예시를 나타낸다.
도17은 실시예들에 따른 2D 비디오/이미지 디코더(2D Video/Image Decoder)의 예시를 나타낸다.
도18은 실시예들에 따른 송신 장치의 동작 흐름도의 예시를 나타낸다.
도19는 실시예들에 따른 수신 장치의 동작 흐름도의 예시를 나타낸다.
도20은 실시예들에 따른 포인트 클라우드 데이터 송수신 방법/장치와 연동 가능한 구조의 예시를 나타낸다.
도21은 실시예들에 따른 복셀화된 서페이스 라이트 필드 시퀀스를 나타낸다.
도22는 실시예들에 따른 2D 포인트 클라우드 데이터를 포인트 별로 액티비티 리젼(activity region ) 정보에 따라 구분하여 표시한 예시를 나타낸다.
도23은 실시예들에 따른 SLF 데이터 셋으로부터 생성된 5개 어트리뷰트 비디오 및 1개 지오메트리 비디오 구성을 나타낸다.
도24는 실시예들에 따른 패치 별 카메라 뷰포인트 선택 방법 및 텍스쳐 비디오 생성하는 예시를 나타낸다.
도25는 실시예들에 따른 텍스쳐 비디오 생성 방법을 나타낸다.
도26은 실시예들에 따른 V3C 비트스트림 구조를 나타낸다.
도27-28은 실시예들에 따른 일반적인 V3C파라미터 세트이다.
도29는 실시예들에 따른 어트리뷰트 정보이다.
도30은 실시예들에 따른 비디오 합성 구조를 나타낸다.
도31-32는 실시예들에 따른 아틀라스 시퀀스 파라미터 세트를 나타낸다.
도33은 실시예들에 다른 아틀라스 프레임 파라미터 세트를 나타낸다.
도34-35는 실시예들에 따른 아틀라스 프레임 타일 정보를 나타낸다.
도36은 실시예들에 따른 아틀라스 어댑테이션 파라미터를 나타낸다.
도37은 실시예들에 따른 패치 정보 데이터를 나타낸다.
도38-39는 실시예들에 따른 패치 데이터 유닛을 나타낸다.
도40은 실시예들에 따른 V-PCC 포인트 클라우드 데이터 송신 장치를 나타낸다.
도41은 실시예들에 따른 다중 오브젝트를 포함하는 SLF 데이터 셋 예시를 나타낸다.
도42-47은 실시예들에 따른 볼륨메트릭 어노테이션 SEI메시지(Volumetric annotation SEI message family syntax)에 포함되는 신 오브젝트 정보 SEI메시지 신택스(Scene object information SEI message syntax )를 나타낸다.
도48은 실시예들에 따른 포인트 클라우드 데이터 송신 방법을 나타낸다.
도49는 실시예들에 따른 포인트 클라우드 데이터 수신 방법을 나타낸다.
실시예들의 바람직한 실시예에 대해 구체적으로 설명하며, 그 예는 첨부된 도면에 나타낸다. 첨부된 도면을 참조한 아래의 상세한 설명은 실시예들의 실시예에 따라 구현될 수 있는 실시예만을 나타내기보다는 실시예들의 바람직한 실시예를 설명하기 위한 것이다. 다음의 상세한 설명은 실시예들에 대한 철저한 이해를 제공하기 위해 세부 사항을 포함한다. 그러나 실시예들이 이러한 세부 사항 없이 실행될 수 있다는 것은 당업자에게 자명하다.
실시예들에서 사용되는 대부분의 용어는 해당 분야에서 널리 사용되는 일반적인 것들에서 선택되지만, 일부 용어는 출원인에 의해 임의로 선택되며 그 의미는 필요에 따라 다음 설명에서 자세히 서술한다. 따라서 실시예들은 용어의 단순한 명칭이나 의미가 아닌 용어의 의도된 의미에 근거하여 이해되어야 한다.
도1은 실시예들에 따른 Point Cloud 콘텐츠 제공을 위한 송신/수신 시스템의 구조의 예시를 나타낸다.
본 문서에서는 사용자에게 VR (Virtual Reality, 가상현실), AR (Augmented Reality, 증강현실), MR (Mixed Reality, 혼합현실), 및 자율 주행 서비스 등의 다양한 서비스를 제공하기 위하여 Point Cloud 콘텐츠를 제공하는 방안을 제공한다. 실시예들에 다른 포인트 클라우드 콘텐츠는 오브젝트를 포인트들로 표현한 데이터를 나타내고, 포인트 클라우드, 포인트 클라우드 데이터, 포인트 클라우드 비디오 데이터, 포인트 클라우드 이미지 데이터 등으로 지칭될 수 있다.
실시예들에 따른 포인트 클라우드 데이터 전송 장치(Transmission device, 10000)는 포인트 클라우드 비디오 획득부(Point Cloud Video Acquisition, 10001), 포인트 클라우드 비디오 인코더(Point Cloud Video Encoder, 10002), 파일/세그먼트 인캡슐레이션부(10003) 및/또는 트랜스미터(Transmitter (or Communication module), 10004)를 포함한다. 실시예들에 따른 전송 장치는 포인트 클라우드 비디오(또는 포인트 클라우드 콘텐트)를 확보하고 처리하여 전송할 수 있다. 실시예들에 따라, 전송 장치는 고정국(fixed station), BTS(base transceiver system), 네트워크, AI(Ariticial Intelligence) 기기 및/또는 시스템, 로봇, AR/VR/XR 기기 및/또는 서버 등을 포함할 수 있다. 또한 실시예들에 따라 전송 장치(10000)는 무선 접속 기술(예, 5G NR(New RAT), LTE(Long Term Evolution))을 이용하여, 기지국 및/또는 다른 무선 기기와 통신을 수행하는 기기, 로봇, 차량, AR/VR/XR 기기, 휴대기기, 가전, IoT(Internet of Thing)기기, AI 기기/서버 등을 포함할 수 있다.
실시예들에 따른 포인트 클라우드 비디오 획득부(Point Cloud Video Acquisition, 10001)는 Point Cloud 비디오의 캡처, 합성 또는 생성 과정 등을 통한 Point Cloud 비디오를 획득한다.
실시예들에 따른 포인트 클라우드 비디오 인코더(Point Cloud Video Encoder, 10002)는 포인트 클라우드 비디오 데이터를 인코딩한다. 실시예들에 따라, 포인트 클라우드 비디오 인코더(10002)는 포인트 클라우드 인코더, 포인트 클라우드 데이터 인코더, 인코더 등으로 지칭될 수 있다. 또한 실시예들에 따른 포인트 클라우드 컴프레션 코딩(인코딩)은 상술한 실시예에 국한되는 것은 아니다. 포인트 클라우드 비디오 인코더는 인코딩된 포인트 클라우드 비디오 데이터를 포함하는 비트스트림을 출력할 수 있다. 비트스트림은 인코딩된 포인트 클라우드 비디오 데이터뿐만 아니라, 포인트 클라우드 비디오 데이터의 인코딩과 관련된 시그널링 정보를 포함할 수 있다.
실시예들에 따른 인코더는 G-PCC (Geometry-based Point Cloud Compression) 인코딩 방식 및/또는 V-PCC(Video-based Point Cloud Compression) 인코딩 방식을 모두 지원할 수 있다. 또한, 인코더는 포인트 클라우드 (포인트 클라우드 데이터 또는 포인트들을 모두 지칭함) 및/또는 포인트 클라우드에 관한 시그널링 데이터를 인코딩할 수 있다. 실시예들에 따른 인코딩의 구체적인 동작은 이하에서 설명한다.
한편, 본 문서에서 사용하는 V-PCC 용어는 비디오 기반 포인트 클라우드 압축(Video-based Point Cloud Compression (V-PCC))을 의미하고, V-PCC 용어는 비쥬얼 볼륨메트릭 비디오 기반 코딩(Visual Volumetric Video-based Coding (V3C))과 동일하고, 서로 상호 보완하여 지칭될 수 있다.
실시예들에 따른 파일/세그먼트 인캡슐레이션부(File/Segment Encapsulation module, 10003)은 포인트 클라우드 데이터를 파일 및/또는 세그먼트 형태로 인캡슐레이션한다. 실시예들에 따른 포인트 클라우드 데이터 송신 방법/장치는 포인트 클라우드 데이터를 파일 및/또는 세그먼트 형태로 전송할 수 있다.
실시예들에 따른 트랜스미터(Transmitter (or Communication module), 10004)는 인코딩된 포인트 클라우드 비디오 데이터를 비트스트림의 형태로 전송한다. 실시예들에 따라 파일 또는 세그먼트는 네트워크를 통해 수신 장치로 전송되거나, 디지털 저장매체(예를 들면 USB, SD, CD, DVD, 블루레이, HDD, SSD 등)에 저장될 수 있다. 실시예들에 따른 트랜스미터는 수신 장치 (또는 리시버(Receiver)와 4G, 5G, 6G 등의 네트워크를 통해 유/무선 통신 가능하다. 또한 트랜스미터는 네트워크 시스템(예를 들면 4G, 5G, 6G 등의 통신 네트워크 시스템)에 따라 필요한 데이터 처리 동작을 수행할 수 있다. 또한 전송 장치는 온 디맨드(On Demand) 방식에 따라 인캡슐레이션된 데이터를 전송할 수도 있다.
실시예들에 따른 포인트 클라우드 데이터 수신 장치(Reception device, 10005)는 리시버(Receiver, 10006), 파일/세그먼트 디캡슐레이션부(10007), 포인트 클라우드 비디오 디코더(Point Cloud Decoder, 10008), 및/또는 렌더러(Renderer, 10009)를 포함한다. 실시예들에 따라 수신 장치는 무선 접속 기술(예, 5G NR(New RAT), LTE(Long Term Evolution))을 이용하여, 기지국 및/또는 다른 무선 기기와 통신을 수행하는 기기, 로봇, 차량, AR/VR/XR 기기, 휴대기기, 가전, IoT(Internet of Thing)기기, AI 기기/서버 등을 포함할 수 있다.
실시예들에 따른 리시버(Receiver, 10006)는 포인트 클라우드 비디오 데이터를 포함하는 비트스트림을 수신한다. 실시예들에 따라 리시버(10006)는 피드백 정보(Feedback Information)을 포인트 클라우드 데이터 전송 장치(10000)에 전송할 수 있다.
파일/세그먼트 디캡슐레이션부(File/Segment Decapsulation module, 10007)은 포인트 클라우드 데이터를 포함하는 파일 및/또는 세그먼트를 디캡슐레이션한다. 실시예들에 따른 디캡슐레이션부는 실시예들에 따른 인캡슐레이션 과정의 역과정을 수행할 수 있다.
포인트 클라우드 비디오 디코더(Point Cloud Decoder, 10007)는 수신된 포인트 클라우드 비디오 데이터를 디코딩한다. 실시예들에 따른 디코더는 실시예들에 따른 인코딩의 역과정을 수행할 수 있다.
렌더러(Renderer, 10007)는 디코딩된 포인트 클라우드 비디오 데이터를 렌더링한다. 실시예들에 따라 렌더러(10007)는 수신단 측에서 획득된 피드백 정보를 포인트 클라우드 비디오 디코더(10006)에 전송할 수 있다. 실시예들에 따른 포인트 클라우드 비디오 데이터는 피드백 정보를 리시버에 전송할 수 있다. 실시예들에 따라 포인트 클라우드 전송 장치가 수신한 피드백 정보는 포인트 클라우드 비디오 인코더에 제공될 수 있다.
도면에 점선으로 표시된 화살표는 수신 장치(10005)에서 획득한 피드백 정보(feedback information)의 전송 경로를 나타낸다. 피드백 정보는 포인트 클라우드 컨텐트를 소비하는 사용자와의 인터랙티비를 반영하기 위한 정보로서, 사용자의 정보(예를 들면 헤드 오리엔테이션 정보), 뷰포트(Viewport) 정보 등)을 포함한다. 특히 포인트 클라우드 콘텐트가 사용자와의 상호작용이 필요한 서비스(예를 들면 자율주행 서비스 등)를 위한 콘텐트인 경우, 피드백 정보는 콘텐트 송신측(예를 들면 전송 장치(10000)) 및/또는 서비스 프로바이더에게 전달될 수 있다. 실시예들에 따라 피드백 정보는 전송 장치(10000) 뿐만 아니라 수신 장치(10005)에서도 사용될 수 있으며, 제공되지 않을 수도 있다.
실시예들에 따른 헤드 오리엔테이션 정보는 사용자의 머리 위치, 방향, 각도, 움직임 등에 대한 정보이다. 실시예들에 따른 수신 장치(10005)는 헤드 오리엔테이션 정보를 기반으로 뷰포트 정보를 계산할 수 있다. 뷰포트 정보는 사용자가 바라보고 있는 포인트 클라우드 비디오의 영역에 대한 정보이다. 시점(viewpoint)은 사용자가 포인트 클라우 비디오를 보고 있는 점으로 뷰포트 영역의 정중앙 지점을 의미할 수 있다. 즉, 뷰포트는 시점을 중심으로 한 영역으로서, 영역의 크기, 형태 등은 FOV(Field Of View) 에 의해 결정될 수 있다. 따라서 수신 장치(10004)는 헤드 오리엔테이션 정보 외에 장치가 지원하는 수직(vertical) 혹은 수평(horizontal) FOV 등을 기반으로 뷰포트 정보를 추출할 수 있다. 또한 수신 장치(10005)는 게이즈 분석 (Gaze Analysis) 등을 수행하여 사용자의 포인트 클라우드 소비 방식, 사용자가 응시하는 포인트 클라우 비디오 영역, 응시 시간 등을 확인한다. 실시예들에 따라 수신 장치(10005)는 게이즈 분석 결과를 포함하는 피드백 정보를 송신 장치(10000)로 전송할 수 있다. 실시예들에 따른 피드백 정보는 렌더링 및/또는 디스플레이 과정에서 획득될 수 있다. 실시예들에 따른 피드백 정보는 수신 장치(10005)에 포함된 하나 또는 그 이상의 센서들에 의해 확보될 수 있다. 또한 실시예들에 따라 피드백 정보는 렌더러(10009) 또는 별도의 외부 엘레멘트(또는 디바이스, 컴포넌트 등)에 의해 확보될 수 있다. 도1의 점선은 렌더러(10009)에서 확보한 피드백 정보의 전달 과정을 나타낸다. 포인트 클라우드 콘텐트 제공 시스템은 피드백 정보를 기반으로 포인트 클라우드 데이터를 처리(인코딩/디코딩)할 수 있다. 따라서 포인트 클라우드 비디오 데이터 디코더(10008)는 피드백 정보를 기반으로 디코딩 동작을 수행할 수 있다. 또한 수신 장치(10005)는 피드백 정보를 전송 장치로 전송할 수 있다. 전송 장치(또는 포인트 클라우드 비디오 데이터 인코더(10002))는 피드백 정보를 기반으로 인코딩 동작을 수행할 수 있다. 따라서 포인트 클라우드 콘텐트 제공 시스템은 모든 포인트 클라우드 데이터를 처리(인코딩/디코딩)하지 않고, 피드백 정보를 기반으로 필요한 데이터(예를 들면 사용자의 헤드 위치에 대응하는 포인트 클라우드 데이터)를 효율적으로 처리하고, 사용자에게 포인트 클라우드 콘텐트를 제공할 수 있다.
실시예들에 따라, 전송 장치(10000)는 인코더, 전송 디바이스, 전송기 등으로 호칭될 수 있으며, 수신 장치(10004)는 디코더, 수신 디바이스, 수신기 등으로 호칭될 수 있다.
실시예들에 따른 도 1 의 포인트 클라우드 콘텐트 제공 시스템에서 처리되는 (획득/인코딩/전송/디코딩/렌더링의 일련의 과정으로 처리되는) 포인트 클라우드 데이터는 포인트 클라우드 콘텐트 데이터 또는 포인트 클라우드 비디오 데이터라고 호칭할 수 있다. 실시예들에 따라 포인트 클라우드 콘텐트 데이터는 포인트 클라우드 데이터와 관련된 메타데이터 내지 시그널링 정보를 포함하는 개념으로 사용될 수 있다.
도 1에 도시된 포인트 클라우드 콘텐트 제공 시스템의 엘리먼트들은 하드웨어, 소프트웨어, 프로세서 및/또는 그것들의 결합등으로 구현될 수 있다.
실시예들은 사용자에게 VR (Virtual Reality, 가상현실), AR (Augmented Reality, 증강현실), MR (Mixed Reality, 혼합현실), 및 자율 주행 서비스 등 다양한 서비스를 제공하기 위하여 포인트 클라우드(Point Cloud) 콘텐츠를 제공할 수 있다.
Point Cloud 콘텐츠 서비스를 제공하기 위하여, 먼저 Point Cloud 비디오가 획득될 수 있다. 획득된 Point Cloud 비디오는 일련의 과정을 거쳐 전송되고, 수신측에서는 수신된 데이터를 다시 원래의 Point Cloud 비디오로 가공하여 렌더링 할 수 있다. 이를 통해 Point Cloud 비디오가 사용자에게 제공될 수 있다. 실시예들은 이러한 일련의 과정을 효과적으로 수행하기 위해 필요한 방안을 제공한다.
Point Cloud 콘텐츠 서비스를 제공하기 위한 전체의 과정(포인트 클라우드 데이터 전송 방법 및/또는 포인트 클라우드 데이터 수신 방법)은 획득 과정, 인코딩 과정, 전송 과정, 디코딩 과정, 렌더링 과정 및/또는 피드백 과정을 포함할 수 있다.
실시예들에 따라 포인트 클라우드 콘텐츠 (또는 포인트 클라우드 데이터)를 제공하는 과정은 포인트 클라우드 컴프레션(Point Cloud Compression) 과정이라고 호칭할 수 있다. 실시예들에 따라 포인트 클라우드 컴프레션 과정은 지오메트리 기반 포인트 클라우드 컴프레션(Geometry-based Point Cloud Compression) 과정을 의미할 수 있다.
실시예들에 따른 포인트 클라우드 데이터 전송 장치 및 포인트 클라우드 데이터 수신 장치의 각 엘리먼트는 하드웨어, 소프트웨어, 프로세서 및/또는 그것들의 결합 등을 의미할 수 있다.
Point Cloud 콘텐츠 서비스를 제공하기 위하여, 먼저 Point Cloud 비디오가 획득될 수 있다. 획득된 Point Cloud 비디오는 일련의 과정을 거쳐 전송되고, 수신측에서는 수신된 데이터를 다시 원래의 Point Cloud 비디오로 가공하여 렌더링 할 수 있다. 이를 통해 Point Cloud 비디오가 사용자에게 제공될 수 있다. 본 발명은 이러한 일련의 과정을 효과적으로 수행하기 위해 필요한 방안을 제공한다.
Point Cloud 콘텐츠 서비스를 제공하기 위한 전체의 과정은 획득 과정, 인코딩 과정, 전송 과정, 디코딩 과정, 렌더링 과정 및/또는 피드백 과정을 포함할 수 있다.
Point Cloud Compression 시스템은 전송 디바이스 및 수신 디바이스를 포함할 수 있다. 전송 디바이스는 Point Cloud 비디오를 인코딩하여 비트스트림을 출력할 수 있으며, 이를 파일 또는 스트리밍 (스트리밍 세그먼트) 형태로 디지털 저장매체 또는 네트워크를 통하여 수신 디바이스로 전달할 수 있다. 디지털 저장 매체는 USB, SD, CD, DVD, 블루레이, HDD, SSD 등 다양한 저장 매체를 포함할 수 있다.
전송 디바이스는 개략적으로 Point Cloud 비디오 획득부, Point Cloud 비디오 인코더, 파일/세그먼트 인캡슐레이션부, 전송부를 포함할 수 있다. 수신 디바이스는 개략적으로 수신부, 파일/세그먼트 디캡슐레이션부, Point Cloud 비디오 디코더 및 렌더러를 포함할 수 있다. 인코더는 Point Cloud 비디오/영상/픽처/프레임 인코딩 장치라고 불릴 수 있고, 디코더는 Point Cloud 비디오/영상/픽처/프레임 디코딩 장치라고 불릴 수 있다. 송신기는 Point Cloud 비디오 인코더에 포함될 수 있다. 수신기는 Point Cloud 비디오 디코더에 포함될 수 있다. 렌더러는 디스플레이부를 포함할 수도 있고, 렌더러 및/또는 디스플레이부는 별개의 디바이스 또는 외부 컴포넌트로 구성될 수도 있다. 전송 디바이스 및 수신 디바이스는 피드백 과정을 위한 별도의 내부 또는 외부의 모듈/유닛/컴포넌트를 더 포함할 수도 있다.
실시예들에 따라 수신 디바이스의 동작은 전송 디바이스 동작의 역과정을 따를 수 있다.
Point Cloud 비디오 획득부는 Point Cloud 비디오의 캡처, 합성 또는 생성 과정 등을 통한 Point Cloud 비디오를 획득하는 과정을 수행할 수 있다. 획득 과정에 의해 다수의 Point들에 대한 3D 위치(x, y, z)/속성 (color, reflectance, transparency 등) 데이터, 예를 들어, PLY(Polygon File format or the Stanford Triangle format) 파일 등이 생성 될 수 있다. 여러 개의 프레임을 갖는 비디오의 경우 하나 이상의 파일들이 획득될 수 있다. 캡처 과정에서 point cloud 관련 메타데이터(예를 들어 캡처와 관련된 메타데이터 등)가 생성될 수 있다.
실시예들에 따른 포인트 클라우드 데이터 송신 장치는 포인트 클라우드 데이터를 인코딩하는 인코더; 및 포인트 클라우드 데이터를 전송하는 트랜스미터; 를 포함할 수 있다. 또한, 포인트 클라우드를 포함하는 비트 스트림의 형태로 전송될 수 있다.
실시예들에 따른 포인트 클라우드 데이터 수신 장치는 포인트 클라우드 데이터를 수신하는 수신부; 포인트 클라우드 데이터를 디코딩하는 디코더; 및 포인트 클라우드 데이터를 렌더링하는 렌더러; 를 포함할 수 있다.
실시예들에 따른 방법/장치는 포인트 클라우드 데이터 송신 장치 및/또는 포인트 클라우드 데이터 수신 장치를 나타낸다.
도2는 실시예들에 따른 포인트 클라우드 데이터 갭쳐의 예시를 나타낸다.
실시예들에 따른 포인트 클라우드 데이터는 카메라 등에 의해 획득될 수 있다. 실시예들에 따른 캡쳐 방법은 예를 들어 인워드-페이싱 및/또는 아웃워드-페이싱이 있을 수 있다.
실시예들에 따른 인워드-페이싱은 포인트 클라우드 데이터의 오브젝트(Object)를 하나 또는 하나 이상의 카메라들이 오브젝트의 바깥에서 안쪽 방향으로 촬영할 수 있다.
실시예들에 따른 아웃워드-페이싱은 포인트 클라우드 데이터의 오브젝트를 하나 또는 하나 이상의 카메라들이 오브젝트의 안쪽에서 바깥 방향으로 촬영할 수 있다. 예를 들어, 실시예들에 따라 카메라는 4개일 수 있다.
실시예들에 따른 포인트 클라우드 데이터 또는 포인트 클라우드 콘텐츠는 다양한 형태의 3D 공간상에 표현되는 객체/환경의 비디오 또는 정지 영상일 수 있다. 실시예들에 따라, 포인트 클라우드 콘텐츠는 객체(오브젝트 등)에 대한 비디오/오디오/이미지 등을 포함할 수 있다.
Point Cloud 콘텐츠 캡쳐를 위해서 깊이(depth)를 획득 할 수 있는 카메라 장비(적외선 패턴 프로젝터와 적외선 카메라의 조합)와 깊이 정보에 대응되는 색상 정보를 추출 할 수 있는 RGB 카메라들의 조합으로 구성될 수 있다. 또는 레이저 펄스를 쏘고 반사되어 돌아오는 시간을 측정하여 반사체의 위치 좌표를 측정하는 레이더 시스템을 이용하는 라이다(LiDAR)를 통해 깊이 정보를 추출할 수 있다. 깊이 정보로부터 3차원 공간상의 점들로 구성된 지오메트리(geometry)의 형태를 추출하고, RGB 정보로부터 각 점의 색상/반사를 표현하는 속성(attribute)을 추출할 수 있다. Point Cloud 콘텐츠는 점들에 대한 위치(x, y, z)와 색상(YCbCr 또는 RGB) 또는 반사율(r) 정보로 구성될 수 있다. Point Cloud 콘텐츠는 외부 환경을 캡쳐하는 아웃워드-페이싱(outward-facing) 방식과, 중심 객체를 캡쳐하는 인워드-페이싱(inward-facing) 방식이 있을 수 있다. VR/AR 환경에서 객체(예-캐릭터, 선수, 물건, 배우 등 핵심이 되는 객체)를 360도로 사용자가 자유롭게 볼 수 있는 Point Cloud 콘텐츠로 구성할 경우, 캡쳐 카메라의 구성은 인워드-페이싱 방식을 사용하게 될 수 있다. 자율 주행과 같이 자동차에서 현재 주변 환경을 Point Cloud 콘텐츠로 구성할 경우, 캡쳐 카메라의 구성은 아웃워드-페이싱 방식을 사용하게 될 수 있다. 여러대의 카메라를 통해 Point Cloud 콘텐츠가 캡쳐 될 수 있기 때문에, 카메라들 사이의 글로벌 공간 좌표계(global coordinate system)를 설정하기 위해 콘텐츠를 캡쳐 하기 전에 카메라의 캘리브레이션 과정이 필요할 수도 있다.
Point Cloud 콘텐츠는 다양한 형태의 3D 공간상에 나타내어지는 객체/환경의 비디오 또는 정지 영상일 수 있다.
그 외에 Point Cloud 콘텐츠의 획득 방법은 캡쳐 된 Point Cloud 비디오를 기반으로 임의의 Point Cloud 비디오가 합성 될 수 있다. 또는 컴퓨터로 생성된 가상의 공간에 대한 Point Cloud 비디오를 제공하고자 하는 경우, 실제 카메라를 통한 캡처가 수행되지 않을 수 있다. 이 경우 단순히 관련 데이터가 생성되는 과정으로 해당 캡처 과정이 갈음될 수 있다.
캡쳐된 Point Cloud 비디오는 콘텐츠의 질을 향상시키기 위한 후처리가 필요할 수 있다. 영상 캡쳐 과정에서 카메라 장비가 제공하는 범위에서 최대/최소 깊이 값을 조정할 수 있지만 그 이후에도 원하지 않는 영역의 points 데이터들이 포함될 수 있어서 원하지 않는 영역(예, 배경)을 제거 한다거나, 또는 연결된 공간을 인식하고 구멍(spatial hole)을 메우는 후처리를 수행할 수 있다. 또한 공간 좌표계를 공유하는 카메라들로부터 추출된 Point Cloud는 캘리브레이션 과정을 통해 획득된 각 카메라의 위치 좌표를 기준으로 각 point들에 대한 글로벌 좌표계로의 변환 과정을 통해 하나의 콘텐츠로 통합될 수 있다. 이를 통해 하나의 넓은 범위의 Point Cloud 콘텐츠를 생성할 수도 있고, 또는 point들의 밀도가 높은 Point Cloud 콘텐츠를 획득할 수도 있다.
Point Cloud 비디오 인코더는 입력 Point Cloud 비디오를 하나 이상의 비디오 스트림으로 인코딩할 수 있다. 하나의 비디오는 다수의 프레임을 포함할 수 있으며, 하나의 프레임은 정지 영상/픽처에 대응될 수 있다. 본 문서에서, Point Cloud 비디오라 함은 Point Cloud 영상/프레임/픽처/비디오/오디오/이미지 등을 포함할 수 있으며, Point Cloud 비디오는 Point Cloud 영상/프레임/픽처와 혼용되어 사용될 수 있다. Point Cloud 비디오 인코더는 Video-based Point Cloud Compression (V-PCC) 절차를 수행할 수 있다. Point Cloud 비디오 인코더는 압축 및 코딩 효율을 위하여 예측, 변환, 양자화, 엔트로피 코딩 등의 일련의 절차를 수행할 수 있다. 인코딩된 데이터(인코딩된 비디오/영상 정보)는 비트스트림(bitstream) 형태로 출력될 수 있다. V-PCC 절차에 기반하는 경우 Point Cloud 비디오 인코더는 Point Cloud 비디오를 후술하는 바와 같이 지오메트리 비디오, 어트리뷰트(attribute) 비디오, 어큐판시(occupancy) 맵 비디오, 그리고 부가 정보(auxiliary information)으로 나누어 인코딩할 수 있다. 지오메트리 비디오는 지오메트리 이미지를 포함할 수 있고, 어트리뷰트(attribute) 비디오는 어트리뷰트 이미지를 포함할 수 있고, 어큐판시(occupancy) 맵 비디오는 어큐판시 맵 이미지를 포함할 수 있다. 부가 정보는 부가 패치 정보(auxiliary patch information)를 포함할 수 있다. 어트리뷰트 비디오/이미지는 텍스쳐 비디오/이미지를 포함할 수 있다.
인캡슐레이션 처리부(file/segment encapsulation module, 10003)는 인코딩된 Point cloud 비디오 데이터 및/또는 Point cloud 비디오 관련 메타데이터를 파일 등의 형태로 인캡슐레이션할 수 있다. 여기서 Point cloud 비디오 관련 메타데이터는 메타데이터 처리부 등으로부터 전달받은 것일 수 있다. 메타데이터 처리부는 point cloud 비디오 인코더에 포함될 수도 있고, 또는 별도의 컴포넌트/모듈로 구성될 수도 있다. 인캡슐레이션 처리부는 해당 데이터들을 ISOBMFF 등의 파일 포맷으로 인캡슐레이션하거나, 기타 DASH 세그먼트 등의 형태로 처리할 수 있다. 인캡슐레이션 처리부는 실시예에 따라 Point cloud 비디오 관련 메타데이터를 파일 포맷 상에 포함시킬 수 있다. Point cloud 비디오 메타데이터는 예를 들어 ISOBMFF 파일 포맷 상의 다양한 레벨의 박스(box)에 포함되거나 파일 내에서 별도의 트랙내의 데이터로 포함될 수 있다. 실시예에 따라, 인캡슐레이션 처리부는 Point cloud 비디오 관련 메타데이터 자체를 파일로 인캡슐레이션할 수 있다. 전송 처리부는 파일 포맷에 따라 인캡슐레이션된 Point cloud 비디오 데이터에 전송을 위한 처리를 가할 수 있다. 전송 처리부는 전송부에 포함될 수도 있고, 또는 별도의 컴포넌트/모듈로 구성될 수도 있다. 전송 처리부는 임의의 전송 프로토콜에 따라 Point cloud 비디오비디오 데이터를 처리할 수 있다. 전송을 위한 처리에는 방송망을 통한 전달을 위한 처리, 브로드밴드를 통한 전달을 위한 처리를 포함할 수 있다. 실시예에 따라 전송 처리부는 Point cloud 비디오 데이터 뿐 아니라, 메타데이터 처리부로부터 Point cloud 비디오관련 메타데이터를 전달받아, 이 것에 전송을 위한 처리를 가할 수도 있다.
전송부(10004)는 비트스트림 형태로 출력된 인코딩된 비디오/영상 정보 또는 데이터를 파일 또는 스트리밍 형태로 디지털 저장매체 또는 네트워크를 통하여 수신 디바이스의 수신부로 전달할 수 있다. 디지털 저장 매체는 USB, SD, CD, DVD, 블루레이, HDD, SSD 등 다양한 저장 매체를 포함할 수 있다. 전송부는 미리 정해진 파일 포멧을 통하여 미디어 파일을 생성하기 위한 엘리먼트를 포함할 수 있고, 방송/통신 네트워크를 통한 전송을 위한 엘레멘트를 포함할 수 있다. 수신부는 비트스트림을 추출하여 디코딩 장치로 전달할 수 있다.
수신부(10003)는 본 발명에 따른 point cloud 비디오 전송 장치가 전송한 point cloud 비디오 데이터를 수신할 수 있다. 전송되는 채널에 따라 수신부는 방송망을 통하여 point cloud 비디오 데이터를 수신할 수도 있고, 브로드밴드를 통하여 point cloud 비디오 데이터를 수신할 수도 있다. 혹은 디지털 저장 매체를 통하여 point cloud 비디오 데이터를 수신할 수도 있다.
수신 처리부는 수신된 point cloud 비디오 데이터에 대해 전송 프로토콜에 따른 처리를 수행할 수 있다. 수신 처리부는 수신부에 포함될 수 있고, 또는 별도의 컴포넌트/모듈로 구성될 수도 있다. 전송측에서 전송을 위한 처리가 수행된 것에 대응되도록, 수신 처리부는 전술한 전송 처리부의 역과정을 수행할 수 있다. 수신 처리부는 획득한 point cloud 비디오 데이터는 디캡슐레이션 처리부로 전달하고, 획득한 point cloud 비디오 관련 메타데이터는 메타데이터 파서로 전달할 수 있다. 수신 처리부가 획득하는 point cloud 비디오 관련 메타데이터는 시그널링 테이블의 형태일 수 있다.
디캡슐레이션 처리부(file/segment decapsulation module, 10007)는 수신 처리부로부터 전달받은 파일 형태의 point cloud 비디오 데이터를 디캡슐레이션할 수 있다. 디캡슐레이션 처리부는 ISOBMFF 등에 따른 파일들을 디캡슐레이션하여, point cloud 비디오 비트스트림 내지 point cloud 비디오 관련 메타데이터(메타데이터 비트스트림)를 획득할 수 있다. 획득된 point cloud 비디오 비트스트림은 point cloud 비디오 디코더로, 획득된 point cloud 비디오 관련 메타데이터(메타데이터 비트스트림)는 메타데이터 처리부로 전달할 수 있다. point cloud 비디오 비트스트림은 메타데이터(메타데이터 비트스트림)를 포함할 수도 있다. 메타데이터 처리부는 point cloud 비디오 디코더에 포함될 수도 있고, 또는 별도의 컴포넌트/모듈로 구성될 수도 있다. 디캡슐레이션 처리부가 획득하는 point cloud 비디오 관련 메타데이터는 파일 포맷 내의 박스 혹은 트랙 형태일 수 있다. 디캡슐레이션 처리부는 필요한 경우 메타데이터 처리부로부터 디캡슐레이션에 필요한 메타데이터를 전달받을 수도 있다. point cloud 비디오 관련 메타데이터는 point cloud 비디오 디코더에 전달되어 point cloud 비디오 디코딩 절차에 사용될 수도 있고, 또는 렌더러에 전달되어 point cloud 비디오 렌더링 절차에 사용될 수도 있다.
Point Cloud 비디오 디코더는 비트스트림을 입력받아 Point Cloud 비디오 인코더의 동작에 대응하는 동작을 수행하여 비디오/영상을 디코딩할 수 있다. 이 경우 Point Cloud 비디오 디코더는 Point Cloud 비디오를 후술하는 바와 같이 지오메트리 비디오, 어트리뷰트(attribute) 비디오, 어큐판시(occupancy) 맵 비디오, 그리고 부가 정보(auxilIary information )으로 나누어 디코딩할 수 있다. 지오메트리 비디오는 지오메트리 이미지를 포함할 수 있고, 어트리뷰트(attribute) 비디오는 어트리뷰트 이미지를 포함할 수 있고, 어큐판시(occupancy) 맵 비디오는 어큐판시 맵 이미지를 포함할 수 있다. 부가 정보는 부가 패치 정보(auxiliary patch information)를 포함할 수 있다. 어트리뷰트 비디오/이미지는 텍스쳐 비디오/이미지를 포함할 수 있다.
디코딩된 지오메트리 이미지와 오큐판시 맵 및 부가 패치 정보를 이용하여 3차원 지오메트리가 복원되며 이후 스무딩 과정을 거칠 수 있다. 스무딩된 3차원 지오메트리에 텍스처 이미지를 이용하여 컬러값을 부여함으로써 컬러 포인트 클라우드 영상/픽처가 복원될 수 있다. 렌더러는 복원된 지오메트리, 컬러 포인트 클라우드 영상/픽처를렌더링할 수 있다. 렌더링된 비디오/영상은 디스플레이부를 통하여 디스플레이될 수 있다. 사용자는 VR/AR 디스플레이 또는 일반 디스플레이 등을 통하여 렌더링 된 결과의 전부 또는 일부 영역을 볼 수 있다.
피드백 과정은 렌더링/디스플레이 과정에서 획득될 수 있는 다양한 피드백 정보들을 송신측으로 전달하거나 수신측의 디코더에 전달하는 과정을 포함할 수 있다. 피드백 과정을 통해 Point Cloud 비디오 소비에 있어 인터랙티비티(interactivity) 가 제공될 수 있다. 실시예에 따라, 피드백 과정에서 헤드 오리엔테이션(Head Orientation) 정보, 사용자가 현재 보고 있는 영역을 나타내는 뷰포트(Viewport) 정보 등이 전달될 수 있다. 실시예에 따라, 사용자는 VR/AR/MR/자율주행 환경 상에 구현된 것들과 상호작용 할 수도 있는데, 이 경우 그 상호작용과 관련된 정보가 피드백 과정에서 송신측 내지 서비스 프로바이더 측으로 전달될 수도 있다. 실시예에 따라 피드백 과정은 수행되지 않을 수도 있다.
헤드 오리엔테이션 정보는 사용자의 머리 위치, 각도, 움직임 등에 대한 정보를 의미할 수 있다. 이 정보를 기반으로 사용자가 현재 Point Cloud 비디오 내에서 보고 있는 영역에 대한 정보, 즉 뷰포트 정보가 계산될 수 있다.
뷰포트 정보는 현재 사용자가 Point Cloud 비디오에서 보고 있는 영역에 대한 정보일 수 있다. 이를 통해 게이즈 분석(Gaze Analysis) 이 수행되어, 사용자가 어떠한 방식으로 Point Cloud 비디오를 소비하는지, Point Cloud 비디오의 어느 영역을 얼마나 응시하는지 등을 확인할 수도 있다. 게이즈 분석은 수신측에서 수행되어 송신측으로 피드백 채널을 통해 전달될 수도 있다. VR/AR/MR 디스플레이 등의 장치는 사용자의 머리 위치/방향, 장치가 지원하는 수직(vertical) 혹은 수평(horizontal) FOV 등에 근거하여 뷰포트 영역을 추출할 수 있다.
실시예에 따라, 전술한 피드백 정보는 송신측으로 전달되는 것 뿐 아니라, 수신측에서 소비될 수도 있다. 즉, 전술한 피드백 정보를 이용하여 수신측의 디코딩, 렌더링 과정 등이 수행될 수 있다. 예를 들어, 헤드 오리엔테이션 정보 및/또는 뷰포트 정보를 이용하여 현재 사용자가 보고 있는 영역에 대한 Point Cloud 비디오만 우선적으로 디코딩 및 렌더링 될 수도 있다.
여기서 뷰포트(viewport) 내지 뷰포트 영역이란, 사용자가 Point Cloud 비디오에서 보고 있는 영역을 의미할 수 있다. 시점(viewpoint) 는 사용자가 Point Cloud 비디오에서 보고 있는 지점으로서, 뷰포트 영역의 정중앙 지점을 의미할 수 있다. 즉, 뷰포트는 시점을 중심으로 한 영역인데, 그 영역이 차지하는 크기 형태 등은 FOV(Field Of View) 에 의해 결정될 수 있다.
이 문서는 상술한 바와 같이 Point Cloud 비디오 압축에 관한 것이다. 예를 들어 이 문서에서 개시된 방법/실시예는 MPEG (Moving Picture Experts Group)의 PCC (point cloud compression or point cloud coding) 표준 또는 차세대 비디오/이미지 코딩 표준에 적용될 수 있다.
이 문서에서 픽처(picture)/프레임(frame)은 일반적으로 특정 시간대의 하나의 영상을 나타내는 단위를 의미할 수 있다.
픽셀(pixel) 또는 펠(pel)은 하나의 픽처(또는 영상)을 구성하는 최소의 단위를 의미할 수 있다. 또한, 픽셀에 대응하는 용어로서 '샘플(sample)'이 사용될 수 있다. 샘플은 일반적으로 픽셀 또는 픽셀의 값을 나타낼 수 있으며, 루마(luma) 성분의 픽셀/픽셀값만을 나타낼 수도 있고, 크로마(chroma) 성분의 픽셀/픽셀 값만을 나타낼 수도 있고, 또는 뎁스(depth) 성분의 픽셀/픽셀값만을 나타낼 수도 있다.
유닛(unit)은 영상 처리의 기본 단위를 나타낼 수 있다. 유닛은 픽처의 특정 영역 및 해당 영역에 관련된 정보 중 적어도 하나를 포함할 수 있다. 유닛은 경우에 따라서 블록(block) 또는 영역(area) 등의 용어와 혼용하여 사용될 수 있다. 일반적인 경우, MxN 블록은 M개의 열과 N개의 행으로 이루어진 샘플들(또는 샘플 어레이) 또는 변환 계수(transform coefficient)들의 집합(또는 어레이)을 포함할 수 있다.
도3은 실시예들에 따른 포인트 클라우드 및 지오메트리, 텍스쳐 이미지의 예시를 나타낸다.
실시예들에 따른 포인트 클라우드는 후술할 도4의 V-PCC 인코딩 프로세스에 입력되어 지오메트리 이미지, 텍스쳐 이미지가 생성될 수 있다. 실시예들에 따라, 포인트 클라우드는 포인트 클라우드 데이터와 동일한 의미로 사용될 수 있다.
도면과 같이, 좌측은 포인트 클라우드로서, 3D 공간 상에 오브젝트가 위치하고, 이를 바운딩 박스 등으로 나타낼 수 있는 포인트 클라우드를 나타낸다. 중간은 지오메트리를 나타내고, 우측은 텍스쳐 이미지(논-패딩)를 나타낸다.
비디오 베이스 포인트 클라우드 컴프레션(Video-based Point Cloud Compression) (V-PCC)는 HEVC, VVC 등의 2D video codec을 기반으로 3차원 point cloud 데이터를 압축하는 방법을 제공할 수 있다. V-PCC 압축 과정에서 다음과 같은 데이터 및 정보들이 생성될 수 있다.
어큐판시 맵(occupancy map): point cloud를 이루는 점들을 patch로 나누어 2D 평면에 맵핑할 때 2D 평면의 해당 위치에 데이터가 존재하는 여부를 0 또는 1의 값으로 알려주는 2진 맵 (binary map) 을 나타낸다. 어큐판시 맵(occupancy map)은 아틀라스에 대응하는 2D어레이를 나타내고, 어큐판시 맵의 값은 아틀라스 내 각 샘플 포지션이 3D포인트에 대응하는지 여부를 나타낼 수 있다.
아틀라스(atlas)는 볼륨매트릭 데이터가 렌더링되는 3D공간 내 3D 바운딩 박스에 대응하는 렉텡귤러 프레임에 위치한 2D 바운딩 박스들 및 그에 관련된 정보의 집합이다.
아틀라스 비트스트림(atlas bitstream)은 아틀라스(atlas)를 구성하는 하나 이상의 아틀라스 프레임(atlas frame)들과 관련 데이터들에 대한 비트스트림이다.
아틀라스 프레임(atlas frame)은 패치(patch)들이 프로젝션된 아틀라스 샘플(atlas sample)들의 2D 직사각형 배열이다.
아틀라스 샘플(atlas sample)은 아틀라스(atlas)와 연관된 패치(patch)들이 프로젝션된 직사각형 프레임의 포지션이다.
아틀라스 프레임(atlas frame)는은 타일(tile)로 분할될 수 있다. 타일은 2D 프레임을 분할하는 단위이다. 즉, 타일은 아틀라스라는 포인트 클라우드 데이터의 시그널링 정보를 분할하는 단위이다.
패치(patch): point cloud를 구성하는 점들의 집합으로, 같은 patch에 속하는 점들은 3차원 공간상에서 서로 인접해 있으며 2D 이미지로의 맵핑 과정에서 6면의 bounding box 평면 중 같은 방향으로 맵핑됨을 나타낸다.
지오메트리 이미지(geometry image): point cloud를 이루는 각 점들의 위치 정보 (geometry)를 patch 단위로 표현하는 depth map 형태의 이미지를 나타낸다. 지오메트리 이미지는1 채널의 픽셀 값으로 구성될 수 있다. 지오메트리(geometry)는 포인트 클라우드 프레임에 연관된 좌표들의 세트를 나타낸다.
텍스쳐 이미지(texture image): point cloud를 이루는 각 점들의 색상 정보를 patch 단위로 표현하는 image를 나타낸다. 텍스쳐 이미지는 복수 채널의 픽셀 값 (e.g. 3채널 R, G, B)으로 구성될 수 있다. 텍스쳐는 어트리뷰트에 포함된다. 실시예들에 따라서, 텍스쳐 및/또는 어트리뷰트는 동일한 대상 및/또는 포함관계로 해석될 수 있다.
오실러리 패치 정보(auxiliary patch info): 개별 patch들로부터 point cloud를 재구성하기 위해 필요한 메타데이터를 나타낸다. 어실러리 패치 인포는 patch의 2D/3D 공간에서의 위치, 크기 등에 대한 정보를 포함할 수 있다.
실시예들에 따른 포인트 클라우드 데이터, 예를 들어 V-PCC 컴포넌트들은 아틀라스, 어큐판시 맵, 지오메트리, 어트리뷰트 등을 포함할 수 있다.
아틀라스(atlas)는 2D바운딩 박스들의 집합을 나타낸다. 패치, 예를 들어, 렉텡귤러 프레임에 프로젝션된 패치들일 수 있다. 또한, 3D공간에서 3D 바운딩 박스에 대응할 수 있고, 포인트 클라우드의 서브세트를 나타낼 수 있다.
어트리뷰트(attribute)는 포인트 클라우드 내 각 포인트와 연관된 scalar 또는 vector를 나타내고, 예를 들어, 컬러(colour), 리플렉턴스(reflectance), 서페이스 노멀(surface normal), 타임 스탬프(time stamps), 머터리얼ID(material ID) 등이 있을 수 있다.
실시예들에 따른 포인트 클라우드 데이터는 V-PCC (Video-based Point Cloud Compression) 방식에 따른 PCC 데이터를 나타낸다. 포인트 클라우드 데이터는 복수의 컴포넌트들을 포함할 수 있다. 예를 들어, 어큐판시 맵, 패치, 지오메트리 및/또는 텍스쳐 등을 포함할 수 있다.
도4는 실시예들에 따른 V-PCC 인코딩 처리의 예시를 나타낸다.
도면은 어큐판시 맵(occupancy map), 지오메트리 이미지(geometry image), 텍스쳐 이미지(texture image), 오실러리 패치 정보(auxiliary patch information)을 생성하고 압축하기 위한 V-PCC encoding process를 도시하여 보여주고 있다. 도4의 V-PCC 인코딩 프로세스는 도1의 포인트 클라우드 비디오 인코더(10002)에 의해 처리될 수 있다. 도4의 각 구성요소는 소프트웨어, 하드웨어, 프로세서 및/또는 그것들의 조합에 의해 수행될 수 있다.
패치 제너레이션(patch generation, 40000) 또는 패치 제너레이터는 포인트 클라우드 프레임(포인트 클라우드 데이터를 포함하는 비트스트림의 형태일 수 있다)을 수신한다. 패치 제너레이션부(40000)는 포인트 클라우드 데이터로부터 패치를 생성한다. 또한, 패치 생성에 관한 정보를 포함하는 패치 인포를 생성한다.
패치 패킹(patch packing, 40001) 또는 패치 패커는 포인트 클라우드 데이터에 대한 패치를 패킹한다. 예를 들어, 하나 또는 하나 이상의 패치들이 패킹될 수 있다. 또한, 패치 패킹에 관한 정보를 포함하는 어큐판시 맵을 생성한다.
지오메트리 이미지 제너레이션(geometry image generation, 40002) 또는 지오메트리 이미지 제너레이터는 포인트 클라우드 데이터, 패치, 및/또는 패킹된 패치에 기반하여 지오메트리 이미지를 생성한다. 지오메트리 이미지는 포인트 클라우드 데이터에 관한 지오메트리를 포함하는 데이터를 말한다.
텍스쳐 이미지 제너레이션(texture image generation, 40003) 또는 텍스쳐 이미지 제너레이터는 포인트 클라우드 데이터, 패치, 및/도는 패킹된 패치에 기반하여 텍스쳐 이미지를 생성한다. 또한, 재구성된(리컨스트럭션된) 지오메트리 이미지를 패치 인포에 기반하여 스무딩(번호)이 스무딩 처리를 하여 생성된 스무딩된 지오메트리에 더 기초하여, 텍스쳐 이미지를 생성할 수 있다.
스무딩(smoothing, 40004) 또는 스무더는 이미지 데이터에 포함된 에러를 완화 또는 제거할 수 있다. 예를 들어, 재구성된 지오메트리 이미지를 패치 인포에 기반하여 데이터 간 에러를 유발할 수 있는 부분을 부드럽게 필터링하여 스무딩된 지오메트리를 생성할 수 있다.
오실러리 패치 인포 컴프레션(auxillary patch info compression, 40005) 또는 오실러리 패치 정보 컴프레서는 패치 생성 과정에서 생성된 패치 정보와 관련된 부가적인 패치 정보를 컴프레션한다. 또한, 컴프레스된 오실러리 패치 인포를 멀티플레서에 전달하고, 지오메트리 이미지 제너레이션(40002)도 오실러리 패치 정보를 이용할 수 있다.
이미지 패딩(image padding, 40006, 40007) 또는 이미지 패더는 지오메트리 이미지 및 텍스쳐 이미지를 각각 패딩할 수 있다. 패딩 데이터가 지오메트리 이미지 및 텍스쳐 이미지에 패딩될 수 있다.
그룹 딜레이션(group dilation, 40008) 또는 그룹 딜라이터는 이미지 패딩과 유사하게, 텍스쳐 이미지에 데이터를 부가할 수 있다. 부가 데이터가 텍스쳐 이미지에 삽입될 수 있다.
비디오 컴프레션(video compression, 40009, 40010, 40011) 또는 비디오 컴프레서는 패딩된 지오메트리 이미지, 패딩된 텍스쳐 이미지 및/또는 어큐판시 맵을 각각 컴프레션할 수 있다. 컴프레션은 지오메트리 정보, 텍스쳐 정보, 어큐판시 정보 등을 인코딩할 수 있다.
엔트로피 컴프레션(entropy compression, 40012) 또는 엔트로피 컴프레서는 어큐판시 맵을 엔트로피 방식에 기반하여 컴프레션(예를 들어, 인코딩)할 수 있다.
실시예들에 따라, 포인트 클라우드 데이터가 로스리스(lossless)한 경우 및/또는 로시(lossy)한 경우에 따라서, 엔트로피 컴프레션 및/또는 비디오 컴프레션이 각각 수행될 수 있다.
멀티플렉서(multiplexer, 40013)는 컴프레스된 지오메트리 이미지, 컴프레스된 텍스쳐 이미지, 컴프레스된 어큐판시 맵을 비트스트림으로 멀티플렉싱한다.
실시예들에 따른 도4의 각 프로세스의 상세한 동작은 다음과 같다.
패치 제너레이션(Patch generation, 40000)
패치 제너레이션(Patch generation) 과정은 포인트 클라우드를 2D 이미지에 맵핑 (mapping)하기 위하여, 맵핑을 수행하는 단위인 patch로 point cloud를 분할하는 과정을 의미한다. Patch generation 과정은 다음과 같이 노멀(normal) 값 계산, 세그멘테이션(segmentation), 패치(patch) 분할의 세 단계로 구분될 수 있다.
도5를 참조하여, 노멀 값 계산 과정을 구체적으로 설명한다.
도5는 실시예들에 따른 서페이스(Surface)의 탄젠트 플렌(tangent plane) 및 노멀 벡터(normal vector)의 예시를 나타낸다.
도5의 서페이스는 도4의 V-PCC 인코딩 프로세스의 패치 제너레이션 과정(40000)에서 다음과 같이 이용된다.
패치 제너레이션 관련하여 노멀(Normal) 계산:
포인트 클라우드를 이루는 각 점(예를 들어, 포인트)들은 고유의 방향을 가지고 있는데 이것은 normal이라는 3차원 vector로 표현된다. K-D tree 등을 이용하여 구해지는 각 점들의 인접점들 (neighbors)을 이용하여, 도면과 같은 point cloud의 surface를 이루는 각 점들의 tangent plane 및 normal vector를 구할 수 있다. 인접점들을 찾는 과정에서의 search range는 사용자에 의해 정의될 수 있다.
탄젠트 플레인(tangent plane): surface의 한 점을 지나면서 surface 위의 곡선에 대한 접선을 완전이 포함하고 있는 평면을 나타낸다.
도6은 실시예들에 따른 포인트 클라우드의 바운딩 박스(bounding box)의 예시를 나타낸다.
실시예들에 따른 방법/장치, 예를 들어, 패치 제너레이션이 포인트 클라우드 데이터로부터 패치를 생성하는 과정에서 바운딩 박스를 이용할 수 있다.
실시예들에 따른 바운딩 박스란, 포인트 클라우드 데이터를 3D 공간 상에서 육면체에 기반하여 분할하는 단위의 박스를 말한다.
바운딩 박스는 포인트 클라우드 데이터의 대상이 되는 오브젝트를 3D 공간 상의 육면체에 기반하여 각 육면체의 평면에 프로젝션하는 과정에서 이용될 수 있다. 바운딩 박스는 도1의 포인트 클라우드 비디오 획득부(10000), 포인트 클라우드 비디오 인코더(10002)에 의해 생성되고 처리될 수 있다. 또한, 바운딩 박스에 기반하여, 도2의 V-PCC 인코딩 프로세스의 패치 제너레이션(40000), 패치 패킹(40001), 지오메트리 이미지 제너레이션(40002), 텍스쳐 이미지 제너레이션(40003)이 수행될 수 있다.
패치 제너레이션 관련하여 세그멘테이션(Segmentation)
세그멘테이션(Segmentation)은 이니셜 세그멘테이션(initial segmentation)과 리파인 세그멘테이션(refine segmentation)의 두 과정으로 이루어 진다.
실시예들에 따른 포인트 클라우드 인코더(10002)는 포인트를 바운딩박스의 한 면에 프로젝션한다. 구체적으로, Point cloud를 이루는 각 점들은 도면과 같이 point cloud를 감싸는 6개의 bounding box의 면들 중 하나의 면에 projection되는데, initial segmentation은 각 점들이 projection될 bounding box의 평면들 중 하나를 결정하는 과정이다.
6개의 각 평면들과 대응되는 normal값인
Figure PCTKR2021011600-appb-img-000001
는 다음과 같이 정의된다.
(1.0, 0.0, 0.0), (0.0, 1.0, 0.0), (0.0, 0.0, 1.0), (-1.0, 0.0, 0.0), (0.0, -1.0, 0.0), (0.0, 0.0, -1.0).
다음의 수식과 같이 앞서 normal 값 계산과정에서 얻은 각 점들의 normal 값(
Figure PCTKR2021011600-appb-img-000002
)과
Figure PCTKR2021011600-appb-img-000003
의 외적 (dot product)이 최대인 면을 해당 면의 projection 평면으로 결정한다. 즉, point의 normal과 가장 유사한 방향의 normal을 갖는 평면이 해당 point 의 projection 평면으로 결정된다.
Figure PCTKR2021011600-appb-img-000004
결정된 평면은 0~5 중 하나의 index 형태의 값 (cluster index) 으로 식별될 수 있다.
Refine segmentation은 앞서 initial segmentation 과정에서 결정된 point cloud를 이루는 각 점의projection 평면을 인접 점들의 projection 평면을 고려하여 개선하는 과정이다. 이 과정에서는 앞서 initial segmentation 과정에서 projection 평면 결정을 위해 고려된 각 포인트의 normal과 bounding box의 각 평면의 normal 값과의 유사 정도를 이루는 score normal과 함께, 현재 점의 projection 평면과 인접 점들의 projection 평면과의 일치 정도를 나타내는 score smooth가 동시에 고려될 수 있다.
Score smooth는 score normal에 대하여 가중치를 부여하여 고려될 수 있으며, 이 때 가중치 값은 사용자에 의해 정의될 수 있다. Refine segmentation은 반복적으로 수행될 수 있으며, 반복 횟수 또한 사용자에 의해 정의될 수 있다.
패치 제너레이션 관련하여 Patch 분할 (segment patches)
Patch 분할은 앞서 initial/refine segmentation 과정에서 얻은 point cloud를 이루는 각 점들의 projection 평면 정보를 바탕으로, 전체 point cloud를 인접한 점들의 집합인 patch로 나누는 과정이다. Patch 분할은 다음과 같은 단계들로 구성될 수 있다.
① K-D tree 등을 이용하여 point cloud를 이루는 각 점들의 인접 점들을 산출한다. 최대 인접점으 개수는 사용자에 의해 정의될 수 있다.
② 인접 점들이 현재의 점과 동일한 평면에 projection 될 경우 (동일한 cluster index 값을 가질 경우) 현재의 점과 해당 인접 점들을 하나의 patch로 추출한다.
③ 추출된 patch의 geometry 값들을 산출한다. 자세한 과정은 이하에서 설명한다.
④ 추출되지 않은 점들이 없어질 때까지 ②④과정을 반복한다.
Patch 분할 과정을 통해 각 patch의 크기 및 patch별 occupancy map, geometry image, texture image 등이 결정된다.
도7은 실시예들에 따른 어큐판시 맵(occupancy map)의 개별 패치(patch) 위치 결정의 예시를 나타낸다.
실시예들에 따른 포인트 클라우드 인코더(10002)는 패치 패킹 및 어큐판시 맵을 생성할 수 있다.
패치 패킹 및 어큐판시 맵 생성(Patch packing & Occupancy map generation, 40001)
본 과정은 앞서 분할된 patch들을 하나의 2D 이미지에 맵핑하기 위해 개별 patch들의 2D 이미지 내에서의 위치를 결정하는 과정이다. Occupancy map은 2D 이미지의 하나로, 해당 위치에 데이터가 존재하는지 여부를 0 또는 1의 값으로 알려주는 바이너리 맵(binary map)이다. Occupancy map은 블록(block)으로 이루어 지며 block의 크기에 따라 그 해상도가 결정될 수 있는데, 일례로 block 크기가 1*1일 경우 픽셀 (pixel) 단위의 해상도를 갖는다. Block의 크기 (occupancy packing block size)는 사용자에 의해 결정될 수 있다.
Occupancy map 내에서 개별 patch의 위치를 결정하는 과정은 다음과 같이 구성될 수 있다.
① 전체 occupancy map의 값들을 모두 0으로 설정한다.
② occupancy map 평면에 존재하는 수평 좌표가 [0, occupancySizeU - patch.sizeU0), 수직 좌표가 [0, occupancySizeV - patch.sizeV0) 범위에 있는 점 (u, v)에 patch를 위치시킨다.
③ patch 평면에 존재하는 수평 좌표가 [0, patch.sizeU0), 수직 좌표가 [0, patch.sizeV0) 범위에 있는 점 (x, y)를 현재 점(포인트)으로 설정한다.
④ 점 (x, y)에 대하여, patch occupancy map의 (x, y) 좌표 값이 1이고 (patch 내 해당 지점에 데이터가 존재하고), 전체 occupancy map의 (u+x, v+y) 좌표 값이 1 (이전 patch에 의해 occupancy map이 채워진 경우) raster order 순으로 (x, y) 위치를 변경하여 ③④의 과정을 반복한다. 그렇지 않을 경우, ⑥의 과정을 수행한다.
⑤ raster order 순으로 (u, v) 위치를 변경하여 ③⑤의 과정을 반복한다.
⑥ (u, v)를 해당 patch의 위치로 결정하고, patch의 occupancy map 데이터를 전체 occupancy map의 해당 부분에 할당(copy)한다.
⑦ 다음 patch에 대하여 ②⑦의 과정을 반복한다.
어큐판시 사이즈U(occupancySizeU): occupancy map의 너비(width)를 나타내며, 단위는 어큐판시 패킹 사이즈 블록(occupancy packing block size) 이다.
어큐판시 사이즈V(occupancySizeV): occupancy map의 높이(height)를 나타내며, 단위는 occupancy packing block size 이다.
패치 사이즈 U0(patch.sizeU0): occupancy map의 width를 나타내며, 단위는 occupancy packing block size 이다.
패치 사이즈 V0(patch.sizeV0): occupancy map의 height를 나타내며, 단위는 occupancy packing block size 이다.
예를 들어, 도7과 같이 어큐판사 패킹 사이즈 블록에 해당하는 박스 내 패치 사이즈를 갖는 패치에 대응하는 박스가 존재하고, 박스 내 포인트(x, y)가 위치할 수 있다.
도8은 실시예들에 따른 노멀(normal), 탄젠트(tangent), 바이탄젠트(bitangent) 축의 관계의 예시를 나타낸다.
실시예들에 따른 포인트 클라우드 인코더(10002)는 지오메트리 이미지를 생성할 수 있다. 지오메트리 이미지란, 포인트 클라우드의 지오메트리 정보를 포함하는 이미지 데이터를 의미한다. 지오메트리 이미지 생성 과정은 도8의 패치의 세 가지 축(노멀, 탄젠트, 바이탄젠트)을 이용할 수 있다.
지오메트리 이미지 생성(Geometry image generation, 40002)
본 과정에서는 개별 patch의 geometry image를 구성하는 depth 값들을 결정하고, 앞서 패치 패킹(patch packing) 과정에서 결정된 patch의 위치를 바탕으로 전체 geometry image를 생성한다. 개별 patch의 geometry image를 구성하는 depth 값들을 결정하는 과정은 다음과 같이 구성될 수 있다.
① 개별 patch의 위치, 크기 관련 파라미터들이 산출된다. 파라미터들은 다음과 같은 정보들을 포함할 수 있다.
normal 축을 나타내는 index: normal은 앞서 patch generation 과정에서 구해지며, tangent 축은 normal과 직각인 축들 중 patch image의 수평(u)축과 일치하는 축이며, bitangent 축은 normal과 직각인 축들 중 patch image의 수직(v)축과 일치하는 축으로, 세 가지 축은 도면과 같이 표현될 수 있다.
도9는 실시예들에 따른 프로젝션 모드의 최소 모드 및 최대 모드의 구성의 예시를 나타낸다.
실시예들에 따른 포인트 클라우드 인코더(10002)는 지오메트리 이미지를 생성하기 위해서 패치에 기반한 프로젝션을 수행할 수 있고, 실시예들에 따른 프로젝션의 모드는 최소 모드 및 최대 모드가 있다.
patch의 3D 공간 좌표: patch를 감싸는 최소 크기의 bounding box를 통해 산출될 수 있다. 예를 들어, patch의 tangent 방향 최소값 (patch 3d shift tangent axis), patch의 bitangent 방향 최소값 (patch 3d shift bitangent axis), patch의 normal 방향 최소값 (patch 3d shift normal axis) 등이 포함될 수 있다.
patch의 2D 크기: patch가 2D 이미지로 패킹될 때의 수평, 수직 방향 크기를 나타낸다. 수평 방향 크기 (patch 2d size u)는 bounding box의 tangent 방향 최대값과 최소값의 차이로, 수직 방향 크기 (patch 2d size v)는 bounding box의 bitangent 방향 최대값과 최소값의 차이로 구해질 수 있다.
② Patch의 projection mode를 결정한다. Projection mode는 최소 모드(min mode)와 최대 모드(max mode) 중 하나일 수 있다. Patch의 geometry 정보는 depth 값으로 표현되는데, patch의 normal 방향으로 patch를 이루는 각 점들을 projection 할 때 depth 값의 최대 값으로 구성되는 이미지와 최소값으로 구성되는 이미지 두 계층(layer)의 이미지들이 생성될 수 있다.
두 계층의 이미지 d0와 d1을 생성함에 있어, min mode일 경우 도면과 같이 최소 depth가 d0에 구성되고, 최소 depth로부터 surface thickness 이내에 존재하는 최대 depth가 d1으로 구성될 수 있다.
예를 들어, 포인트 클라우드가 도면과 같이 2D에 위치하는 경우, 복수의 포인트들을 포함하는 복수의 패치들이 있을 수 있다. 도면과 같이 같은 스타일의 음영으로 표시된 포인트들이 동일한 패치에 속할 수 있음을 나타낸다. 빈 칸으로 표시된 포인트들의 패치를 프로젝션하는 과정을 도면이 나타낸다.
빈 칸으로 표시된 포인트들을 좌측/우측으로 프로젝션하는 경우, 좌측을 기준으로 depth를 0, 1, 2,..6, 7, 8, 9 와 같이 1씩 증가하면서 우측으로 포인트들의 depth산출을 위한 숫자를 표기할 수 있다.
프로젝션 모드(Projection mode)는 사용자 정의에 의해 모든 point cloud에 동일한 방법이 적용되거나, frame 또는 patch 별로 다르게 적용될 수 있다. Frame 또는 patch 별로 다른 projection mode가 적용될 경우, 압축 효율을 높이거나 소실 점 (missed point)을 최소화 할 수 있는 projection mode가 적응적으로 선택될 수 있다.
③ 개별 점들의 depth 값을 산출한다.
최소 모드(Min mode)일 경우 각 점의 normal 축 최소값에 patch의 normal 방향 최소값 (patch 3d shift normal axis)에서 ①의 과정에서 산출된 patch의 normal 방향 최소값 (patch 3d shift normal axis)을 뺀 값인 depth0로 d0 이미지를 구성한다. 동일 위치에 depth0와 surface thickness 이내의 범위에 또 다른 depth 값이 존재할 경우, 이 값을 depth1으로 설정한다. 존재하지 않을 경우 depth0의 값을 depth1에도 할당한다. Depth1 값으로 d1 이미지를 구성한다.
예를 들어, d0의 포인트들의 depth를 결정함에 있어 최소값이 산출될 수 있다(4 2 4 4 0 6 0 0 9 9 0 8 0). 그리고, d1의 포인트들의 depth를 결정함에 있어 두 개 이상의 포인트들 중 큰 값이 산출되거나, 하나의 포인트만 있는 경우 그 값이 산출될 수 있다(4 4 4 4 6 6 6 8 9 9 8 8 9). 또한, 패치의 포인트들이 부호화되고, 재구성(reconstruct) 되는 과정에서 일부 포인트가 손실될 수 있다(예를 들어, 도면은 8개의 포인트가 손실되었다).
맥스 모드(Max mode)일 경우 각 점의 normal 축 최대값에 patch의 normal 방향 최소값 (patch 3d shift normal axis)에서 ①의 과정에서 산출된 patch의 normal 방향 최소값 (patch 3d shift normal axis)을 뺀 값인 depth0로 d0 이미지를 구성한다. 동일 위치에 depth0와 surface thickness 이내의 범위에 또 다른 depth 값이 존재할 경우, 이 값을 depth1으로 설정한다. 존재하지 않을 경우 depth0의 값을 depth1에도 할당한다. Depth1 값으로 d1 이미지를 구성한다.
예를 들어, d0의 포인트들의 depth를 결정함에 있어 최대값이 산출될 수 있다(4 4 4 4 6 6 6 8 9 9 8 8 9). 그리고, d1의 포인트들의 depth를 결정함에 있어 두 개 이상의 포인트들 중 작은 값이 산출되거나, 하나의 포인트만 있는 경우 그 값이 산출 될 수 있다(4 2 4 4 5 6 0 6 9 9 0 8 0). 또한, 패치의 포인트들이 부호화되고, 재구성(reconstruct) 되는 과정에서 일부 포인트가 손실될 수 있다(예를 들어, 도면은 6개의 포인트가 손실되었다).
위와 같은 과정을 통해 생성된 개별 patch의 geometry image를 앞서 patch packing 과정에서 결정된 patch의 위치 정보를 이용하여 전체 geometry image에 배치시킴으로써 전체 geometry image를 생성할 수 있다.
생성된 전체 geometry image의 d1 계층은 여러 가지 방법으로 부호화 될 수 있다. 첫 번째는 앞서 생성한 d1 이미지의 depth값들을 그대로 부호화 (absolute d1 method)하는 방법이다. 두 번째는 앞서 생성한 d1 이미지의 depth값과 d0 이미지의 depth값이 차이 값을 부호화 (differential method)하는 방법이다.
이와 같은 d0, d1 두 계층의 depth 값을 이용한 부호화 방법은 두 depth 사이에 또 다른 점들이 존재할 경우 해당 점의 geometry 정보를 부호화 과정에서 잃어버리기 때문에, 무손실 압축 (lossless coding)을 위해 Enhanced-Delta-Depth (EDD) code를 이용할 수도 있다.
도10을 참조하여, EDD code를 구체적으로 설명한다.
도10은 실시예들에 따른 EDD 코드의 예시를 나타낸다.
포인트 클라우드 인코더(10002) 및/또는 V-PCC 인코딩의 일부/전체 프로세스(예를 들어, 비디오 컴프레스(40009)) 등은 EOD코드에 기반하여 포인트들의 지오메트리 정보를 인코딩할 수 있다.
EDD code는 도면과 같이, d1을 포함하여 surface thickness 범위 내의 모든 점들의 위치를 이진으로 부호화 하는 방법이다. 일례로 도면의 좌측에서 두 번째 열에 포함되는 점들의 경우, D0 위쪽으로 첫 번째, 네 번째 위치에 점들이 존재하고, 두 번째와 세 번째 위치는 비어있기 때문에 0b1001 (=9)의 EDD code로 표현될 수 있다. D0와 함께 EDD code를 부호화하여 보내 주면 수신단에서는 모든 점들의 geometry 정보를 손실 없이 복원할 수 있게 된다.
예를 들어, 기준 포인트 위로 포인트가 존재하면 1이고 포인트가 존재하지 않으면 0이 되어 4개의 비트들에 기반하여 코드가 표현될 수 있다.
스무딩(Smoothing, 40004)
스무딩(Smoothing)은 압축 과정에서 발생하는 화질의 열화로 인해 patch 경계면에서 발생할 수 있는 불연속성을 제거하기 위한 작업이며, 포인트 클라우드 인코더 또는 스무더에 의해 수행될 수 있다.
① geometry image로부터 point cloud를 재생성(reconstruction)한다. 본 과정은 앞서 설명한 geometry image 생성의 역과정이라고 할 수 있다. 예를 들어, 인코딩의 역과정이 리컨스트럭션일 수 있다.
② K-D tree 등을 이용하여 재생성된 point cloud를 구성하는 각 점들의 인접점들을 산출한다.
③ 각 점들에 대하여, 해당 점이 patch 경계면에 위치하는지를 판단한다. 일례로 현재 점과 다른 projection 평면 (cluster index)을 갖는 인접점이 존재할 경우, 해당 점은 patch 경계면에 위치한다고 판단할 수 있다.
④ patch 경계면에 존재할 경우, 해당 점을 인접점들의 무게중심 (인접점들의 평균 x, y, z 좌표에 위치)으로 이동시킨다. 즉, geometry 값을 변경시킨다. 그렇지 않을 경위 이전 geometry 값을 유지한다.
도11은 실시예들에 따른 인접점들의 컬러(color) 값들을 이용한 리컬러링(recoloring)의 예시를 나타낸다.
실시예들에 따른 포인트 클라우드 인코더 또는 텍스쳐 이미지 제너레이터(40003)은 리컬러리링에 기반하여 텍스쳐 이미지를 생성할 수 있다.
텍스쳐 이미지 생성(Texture image generation, 40003)
Texture image 생성 과정은 앞서 설명한 geometry image 생성 과정과 유사하게, 개별 patch의 texture image 생성하고, 이들은 결정된 위치에 배치하여 전체 texture image를 생성하는 과정으로 구성된다. 다만 개별 patch의 texture image를 생성하는 과정에 있어서 geometry 생성을 위한 depth 값을 대신하여 해당 위치에 대응되는 point cloud를 구성하는 점의 color 값 (e.g. R, G, B)을 갖는 image가 생성된다.
Point cloud를 구성하는 각 점의 color 값을 구하는 과정에 있어서 앞서 smoothing 과정을 거친 geometry가 사용될 수 있다. Smoothing된 point cloud는 원본 point cloud에서 일부 점들의 위치가 이동된 상태일 수 있으므로, 변경된 위치에 적합한 color를 찾아내는 리컬러링(recoloring)과정이 필요할 수 있다. Recoloring은 인접점들의 color 값들을 이용하여 수행될 수 있다. 일례로, 도면과 같이 새로운 color값은 최인접점의 color값과 인접점들의 color값들을 고려하여 산출될 수 있다.
예를 들어, 도면을 참조하면, 리컬러링은 포인트에 대한 가장 가까운 오리지날 포인트들의 어트리뷰트 정보의 평균 및/또는 포인트에 대한 가장 가까운 오리지날 위치의 어트리뷰트 정보의 평균에 기반하여 변경된 위치의 적합한 컬러값을 산출할 수 있다.
Texture image 또한 d0/d1의 두 계층으로 생성되는 geometry image와 같이 t0/t1의 두 개의 계층 으로 생성될 수 있다.
오실러리 패치 인포 컴프레션(Auxiliary patch info compression, 40005)
실시예들에 따른 포인트 클라우드 인코더 또는 오실러리 패치 정보 컴프레서는 오실러리 패치 정보(포인트 클라우드에 관한 부가적인 정보)를 컴프레션할 수 있다.
오실러리 패치 정보 컴프레서는 앞서 설명한 patch generation, patch packing, geometry generation 과정 등에서 생성된 부가 patch 정보들을 압축(컴프레스)한다. 부가 patch 정보에는 다음과 같은 파라미터들이 포함될 수 있다:
프로젝션(projection) 평면 (normal)을 식별하는 인덱스 (클러스터 인덱스, cluster index)
패치의 3D 공간 위치: 패치의 탄젠트 방향 최소값 (patch 3d shift tangent axis), 패치의 바이탄젠트 방향 최소값 (patch 3d shift bitangent axis), 패치의 노멀 방향 최소값 (patch 3d shift normal axis)
패치의 2D 공간 위치, 크기: 수평 방향 크기 (patch 2d size u), 수직 방향 크기 (patch 2d size v), 수평 방향 최소값 (patch 2d shift u), 수직 방향 최소값 (patch 2d shift u)
각 블록과 패치의 맵핑 정보: candidate index (위의 patch의 2D 공간 위치, 크기 정보를 기반으로 patch를 순서대로 위치시켰을 때, 한 block에 중복으로 복수 patch가 맵핑될 수 있음. 이때 맵핑되는 patch들이 candidate list를 구성하며, 이 list 중 몇 번째 patch의 data가 해당 block에 존재하는지를 나타내는 index), local patch index (frame에 존재하는 전체 patch들 중 하나를 가리키는 index). Table X는 candidate list와 local patch index를 이용한 block과 patch match 과정을 나타내는 pseudo code이다.
candidate list의 최대 개수는 사용자에 의해 정의될 수 있다.
for(i=0; i<BlockCount; i++) {
if(candidatePatches[i].size() == 1) {
blockToPatch[i] = candidatePatches[i][0]
} else {
candidate_index
if(candidate_index == max_candidate_count){
blockToPatch[i] = local_patch_index
} else {
blockToPatch[i] = candidatePatches[i][candidate_index]
}
}
}
도12는 실시예들에 따른 푸쉬-풀 백그라운드 필링(push-pull background filling)의 예시를 나타낸다.
이미지 패딩 및 그룹 딜레이션(Image padding and group dilation, 40006, 40007, 40008)
실시예들에 따른 이미지 패더는 푸쉬-풀 백그라운드 필링 방식에 기반하여 패치 영역 외의 공간을 의미 없는 부가적인 데이터로 채울 수 있다.
이미지 패딩(Image padding)은 압축 효율 향상을 목적으로 patch 영역 이외의 공간을 의미 없는 데이터로 채우는 과정이다. Image padding을 위해 patch 내부의 경계면 쪽에 해당하는 열 또는 행의 픽셀 값들이 복사되어 빈 공간을 채우는 방법이 사용될 수 있다. 또는 도면과 같이, padding 되지 않은 이미지를 단계적으로 해상도를 줄이고, 다시 해상도를 늘리는 과정에서 낮은 해상도의 이미지로부터 온 픽셀 값들로 빈 공간을 채우는 push-pull background filling 방법이 사용될 수도 있다.
그룹 딜레이션은 d0/d1, t0/t1 두 계층으로 이루어진 geometry, texture image의 빈 공간을 채우는 방법으로, 앞서 image padding을 통해 산출된 두 계층 빈 공간의 값들을, 두 계층의 동일 위치에 대한 값의 평균값으로 채우는 과정이다.
도13은 실시예들에 따른 4*4 크기의 블록(block)에 대해 가능한 트라버설 오더(traversal order)의 예시를 나타낸다.
오큐판시 맵 컴프레션(Occupancy map compression, 40012, 40011)
실시예들에 따른 오규판시 맵 컴프레서는 앞서 생성된 occupancy map을 압축할 수 있다. 구체적으로, 손실 (lossy) 압축을 위한 비디오 컴프레션과 무손실 (lossless) 압축을 위한 엔트로피 컴프레션, 두 가지 방법이 존재할 수 있다. 비디오 컴프레션은 이하에서 설명한다.
엔트로피 컴프레션(Entropy compression) 과정은 다음과 같은 과정으로 수행될 수 있다.
① occupancy map을 구성하는 각 block에 대하여, block이 모두 채워진 경우 1을 부호화 하고 다음 block에 대해 동일 과정을 반복한다. 그렇지 않은 경우 0을 부호화하고, ②⑤의 과정을 수행한다. .
② block의 채워진 pixel들에 대해 run-length coding을 수행하기 위한 best traversal order를 결정한다. 도면은 4*4 크기의 block에 대해 가능한 4가지 traversal order를 일례로 보여주고 있다.
도14는 실시예들에 따른 베스트 트라버설 오더의 예시를 나타낸다.
상술한 바와 같이 실시예들에 따른 엔트포리 컴프레서는 도면과 같이 트라버설 오더 방식에 기반하여 블록을 코딩(부호화)할 수 있다.
예를 들어, 가능한 traversal order들 중 최소의 run 개수를 갖는 베스트 트라버설 오더(best traversal order)를 선택하여 그 인덱스를 부호화 한다. 일례로 도면은 앞선 도13의 세 번째 traversal order를 선택할 경우이며, 이 경우 run의 개수가 2로 최소화될 수 있으므로 이를 베스트 트라버설 오더로 선택할 수 있다.
이때 run의 개수를 부호화 한다. 도14의 예에서는 2개의 run이 존재하므로 2가 부호화 된다.
④ 첫 번째 run의 occupancy를 부호화 한다. 도14의 예에서는 첫 번째 run이 채워지지 않은 픽셀들에 해당하므로 0이 부호화된다.
⑤ 개별 run에 대한 (run의 개수만큼의) length를 부호화 한다. 도14의 예에서는 첫 번째 run과 두 번째 run의 length인 6과 10이 순차적으로 부호화된다.
비디오 컴프레션(Video compression, 40009, 40010, 40011)
실시예들에 따른 비디오 컴프레서는 HEVC, VVC 등의 2D video codec 등을 이용하여, 앞서 설명한 과정으로 생성된 geometry image, texture image, occupancy map image 등의 시퀀스를 부호화한다.
도15는 실시예들에 따른 2D 비디오/이미지 인코더(2D video/image Encoder)의 예시를 나타낸다.
도면은 상술한 비디오 컴프레션(Video compression, 40009, 40010, 40011) 또는 비디오 컴프레서의 실시예로서, 비디오/영상 신호의 인코딩이 수행되는 2D 비디오/이미지 인코더(15000)의 개략적인 블록도를 나타낸다. 2D 비디오/이미지 인코더(15000)는 상술한 포인트 클라우드 비디오 인코더에 포함될 수 있고, 또는 내/외부 컴포넌트로 구성될 수도 있다. 도15의 각 구성요소는 소프트웨어, 하드웨어, 프로세서 및/또는그것들의 조합에 대응할 수 있다.
여기서 입력 영상은 상술한 geometry image, texture image (attribute(s) image), occupancy map image 등을 포함할 수 있다. 포인트 클라우드 비디오 인코더의 출력 bitstream (즉, point cloud video/image bitstream)은 각 입력 영상(geometry image, texture image (attribute(s) image), occupancy map image 등)에 대한 출력 비트스트림들을 포함할 수 있다.
인터 예측부(15090) 및 인트라 예측부(15100)를 합쳐서 예측부라고 불릴 수 있다. 즉, 예측부는 인터 예측부(15090) 및 인트라 예측부(15100)를 포함할 수 있다. 변환부(15030), 양자화부(15040), 역양자화부(15050), 역변환부(15060)는 레지듀얼(residual) 처리부에 포함될 수 있다. 레지듀얼 처리부는 감산부(15020)를 더 포함할 수도 있다. 상술한 영상 분할부(15010), 감산부(15020), 변환부(15030), 양자화부(15040), 역양자화부(),), 역변환부(15060), 가산부(155), 필터링부(15070), 인터 예측부(15090), 인트라 예측부(15100) 및 엔트로피 인코딩부(15110)는 실시예에 따라 하나의 하드웨어 컴포넌트(예를 들어 인코더 또는 프로세서)에 의하여 구성될 수 있다. 또한 메모리(15080)는 DPB(decoded picture buffer)를 포함할 수 있고, 디지털 저장 매체에 의하여 구성될 수도 있다.
영상 분할부(15010)는 인코딩 장치(15000)에 입력된 입력 영상(또는, 픽쳐, 프레임)를 하나 이상의 처리 유닛(processing unit)으로 분할할 수 있다. 일 예로, 처리 유닛은 코딩 유닛(coding unit, CU)이라고 불릴 수 있다. 이 경우 코딩 유닛은 코딩 트리 유닛(coding tree unit, CTU) 또는 최대 코딩 유닛(largest coding unit, LCU)으로부터 QTBT (Quad-tree binary-tree) 구조에 따라 재귀적으로(recursively) 분할될 수 있다. 예를 들어, 하나의 코딩 유닛은 쿼드 트리 구조 및/또는 바이너리 트리 구조를 기반으로 하위(deeper) 뎁스의 복수의 코딩 유닛들로 분할될 수 있다. 이 경우 예를 들어 쿼드 트리 구조가 먼저 적용되고 바이너리 트리 구조가 나중에 적용될 수 있다. 또는 바이너리 트리 구조가 먼저 적용될 수도 있다. 더 이상 분할되지 않는 최종 코딩 유닛을 기반으로 본 발명에 따른 코딩 절차가 수행될 수 있다. 이 경우 영상 특성에 따른 코딩 효율 등을 기반으로, 최대 코딩 유닛이 바로 최종 코딩 유닛으로 사용될 수 있고, 또는 필요에 따라 코딩 유닛은 재귀적으로(recursively) 보다 하위 뎁스의 코딩 유닛들로 분할되어 최적의 사이즈의 코딩 유닛이 최종 코딩 유닛으로 사용될 수 있다. 여기서 코딩 절차라 함은 후술하는 예측, 변환, 및 복원 등의 절차를 포함할 수 있다. 다른 예로, 처리 유닛은 예측 유닛(PU: Prediction Unit) 또는 변환 유닛(TU: Transform Unit)을 더 포함할 수 있다. 이 경우 예측 유닛 및 변환 유닛은 각각 상술한 최종 코딩 유닛으로부터 분할 또는 파티셔닝될 수 있다. 예측 유닛은 샘플 예측의 단위일 수 있고, 변환 유닛은 변환 계수를 유도하는 단위 및/또는 변환 계수로부터 레지듀얼 신호(residual signal)를 유도하는 단위일 수 있다.
유닛은 경우에 따라서 블록(block) 또는 영역(area) 등의 용어와 혼용하여 사용될 수 있다. 일반적인 경우, MxN 블록은 M개의 열과 N개의 행으로 이루어진 샘플들 또는 변환 계수(transform coefficient)들의 집합을 나타낼 수 있다. 샘플은 일반적으로 픽셀 또는 픽셀의 값을 나타낼 수 있으며, 휘도(luma) 성분의 픽셀/픽셀값만을 나타낼 수도 있고, 채도(chroma) 성분의 픽셀/픽셀 값만을 나타낼 수도 있다. 샘플은 하나의 픽처(또는 영상)을 픽셀(pixel) 또는 펠(pel)에 대응하는 용어로서 사용될 수 있다.
인코딩 장치(15000)는 입력 영상 신호(원본 블록, 원본 샘플 어레이)에서 인터 예측부(15090) 또는 인트라 예측부(15100)로부터 출력된 예측 신호(예측된 블록, 예측 샘플 어레이)를 감산하여 레지듀얼 신호(residual signal, 잔여 블록, 잔여 샘플 어레이)를 생성할 수 있고, 생성된 레지듀얼 신호는 변환부(15030)로 전송된다. 이 경우 도시된 바와 같이 인코더(15000) 내에서 입력 영상 신호(원본 블록, 원본 샘플 어레이)에서 예측 신호(예측 블록, 예측 샘플 어레이)를 감산하는 유닛은 감산부(15020)라고 불릴 수 있다. 예측부는 처리 대상 블록(이하, 현재 블록이라 함)에 대한 예측을 수행하고, 현재 블록에 대한 예측 샘플들을 포함하는 예측된 블록(predicted block)을 생성할 수 있다. 예측부는 현재 블록 또는 CU 단위로 인트라 예측이 적용되는지 또는 인터 예측이 적용되는지 결정할 수 있다. 예측부는 각 예측모드에 대한 설명에서 후술하는 바와 같이 예측 모드 정보 등 예측에 관한 다양한 정보를 생성하여 엔트로피 인코딩부(15110)로 전달할 수 있다. 예측에 관한 정보는 엔트로피 인코딩부(15110)에서 인코딩되어 비트스트림 형태로 출력될 수 있다.
인트라 예측부(15100)는 현재 픽처 내의 샘플들을 참조하여 현재 블록을 예측할 수 있다. 참조되는 샘플들은 예측 모드에 따라 현재 블록의 주변(neighbor)에 위치할 수 있고, 또는 떨어져서 위치할 수도 있다. 인트라 예측에서 예측 모드들은 복수의 비방향성 모드와 복수의 방향성 모드를 포함할 수 있다. 비방향성 모드는 예를 들어 DC 모드 및 플래너 모드(Planar 모드)를 포함할 수 있다. 방향성 모드는 예측 방향의 세밀한 정도에 따라 예를 들어 33개의 방향성 예측 모드 또는 65개의 방향성 예측 모드를 포함할 수 있다. 다만, 이는 예시로서 설정에 따라 그 이상 또는 그 이하의 개수의 방향성 예측 모드들이 사용될 수 있다. 인트라 예측부(15100)는 주변 블록에 적용된 예측 모드를 이용하여, 현재 블록에 적용되는 예측 모드를 결정할 수도 있다.
인터 예측부(15090)는 참조 픽처 상에서 움직임 벡터에 의해 특정되는 참조 블록(참조 샘플 어레이)을 기반으로, 현재 블록에 대한 예측된 블록을 유도할 수 있다. 이때, 인터 예측 모드에서 전송되는 움직임 정보의 양을 줄이기 위해 주변 블록과 현재 블록 간의 움직임 정보의 상관성에 기초하여 움직임 정보를 블록, 서브블록 또는 샘플 단위로 예측할 수 있다. 움직임 정보는 움직임 벡터 및 참조 픽처 인덱스를 포함할 수 있다. 움직임 정보는 인터 예측 방향(L0 예측, L1 예측, Bi 예측 등) 정보를 더 포함할 수 있다. 인터 예측의 경우에, 주변 블록은 현재 픽처 내에 존재하는 공간적 주변 블록(spatial neighboring block)과 참조 픽처에 존재하는 시간적 주변 블록(temporal neighboring block)을 포함할 수 있다. 참조 블록을 포함하는 참조 픽처와 시간적 주변 블록을 포함하는 참조 픽처는 동일할 수도 있고, 다를 수도 있다. 시간적 주변 블록은 동일 위치 참조 블록(collocated reference block), 동일 위치 CU(colCU) 등의 이름으로 불릴 수 있으며, 시간적 주변 블록을 포함하는 참조 픽처는 동일 위치 픽처(collocated picture, colPic)라고 불릴 수도 있다. 예를 들어, 인터 예측부(15090)는 주변 블록들을 기반으로 움직임 정보 후보 리스트를 구성하고, 현재 블록의 움직임 벡터 및/또는 참조 픽처 인덱스를 도출하기 위하여 어떤 후보가 사용되는지를 지시하는 정보를 생성할 수 있다. 다양한 예측 모드를 기반으로 인터 예측이 수행될 수 있으며, 예를 들어 스킵 모드와 머지 모드의 경우에, 인터 예측부(15090)는 주변 블록의 움직임 정보를 현재 블록의 움직임 정보로 이용할 수 있다. 스킵 모드의 경우, 머지 모드와 달리 레지듀얼 신호가 전송되지 않을 수 있다. 움직임 정보 예측(motion vector prediction, MVP) 모드의 경우, 주변 블록의 움직임 벡터를 움직임 벡터 예측자(motion vector predictor)로 이용하고, 움직임 벡터 차분(motion vector difference)을 시그널링함으로써 현재 블록의 움직임 벡터를 지시할 수 있다.
인터 예측부(15090) 인트라 예측부(15100)를 통해 생성된 예측 신호는 복원 신호를 생성하기 위해 이용되거나 레지듀얼 신호를 생성하기 위해 이용될 수 있다.
변환부(15030)는 레지듀얼 신호에 변환 기법을 적용하여 변환 계수들(transform coefficients)를 생성할 수 있다. 예를 들어, 변환 기법은 DCT(Discrete Cosine Transform), DST(Discrete Sine Transform), KLT(Karhunen-Loeve Transform), GBT(Graph-Based Transform), 또는 CNT(Conditionally Non-linear Transform) 중 적어도 하나를 포함할 수 있다. 여기서, GBT는 픽셀 간의 관계 정보를 그래프로 표현한다고 할 때 이 그래프로부터 얻어진 변환을 의미한다. CNT는 이전에 복원된 모든 픽셀(all previously reconstructed pixel)를 이용하여 예측 신호를 생성하고 그에 기초하여 획득되는 변환을 의미한다. 또한, 변환 과정은 정사각형의 동일한 크기를 갖는 픽셀 블록에 적용될 수도 있고, 정사각형이 아닌 가변 크기의 블록에도 적용될 수 있다.
양자화부(15040)는 변환 계수들을 양자화하여 엔트로피 인코딩부(15110)로 전송되고, 엔트로피 인코딩부(15110)는 양자화된 신호(양자화된 변환 계수들에 관한 정보)를 인코딩하여 비트스트림으로 출력할 수 있다. 양자화된 변환 계수들에 관한 정보는 레지듀얼 정보라고 불릴 수 있다. 양자화부(15040)는 계수 스캔 순서(scan order)를 기반으로 블록 형태의 양자화된 변환 계수들을 1차원 벡터 형태로 재정렬할 수 있고, 1차원 벡터 형태의 양자화된 변환 계수들을 기반으로 양자화된 변환 계수들에 관한 정보를 생성할 수도 있다. 엔트로피 인코딩부(15110)는 예를 들어 지수 골롬(exponential Golomb), CAVLC(context-adaptive variable length coding), CABAC(context-adaptive binary arithmetic coding) 등과 같은 다양한 인코딩 방법을 수행할 수 있다. 엔트로피 인코딩부(15110)는 양자화된 변환 계수들 외 비디오/이미지 복원에 필요한 정보들(예컨대 신택스 요소들(syntax elements)의 값 등)을 함께 또는 별도로 인코딩할 수도 있다. 인코딩된 정보(ex. 인코딩된 비디오/영상 정보)는 비트스트림 형태로 NAL(network abstraction layer) 유닛 단위로 전송 또는 저장될 수 있다. 비트스트림은 네트워크를 통하여 전송될 수 있고, 또는 디지털 저장매체에 저장될 수 있다. 여기서 네트워크는 방송망 및/또는 통신망 등을 포함할 수 있고, 디지털 저장매체는 USB, SD, CD, DVD, 블루레이, HDD, SSD 등 다양한 저장매체를 포함할 수 있다. 엔트로피 인코딩부(15110)로부터 출력된 신호는 전송하는 전송부(미도시) 및/또는 저장하는 저장부(미도시)가 인코딩 장치(15000)의 내/외부 엘리먼트로서 구성될 수 있고, 또는 전송부는 엔트로피 인코딩부(15110)에 포함될 수도 있다.
양자화부(15040)로부터 출력된 양자화된 변환 계수들은 예측 신호를 생성하기 위해 이용될 수 있다. 예를 들어, 양자화된 변환 계수들에 역양자화부(15040) 및 역변환부(15060)를 통해 역양자화 및 역변환을 적용함으로써 레지듀얼 신호(레지듀얼 블록 or 레지듀얼 샘플들)를 복원할 수 있다. 가산부(155)는 복원된 레지듀얼 신호를 인터 예측부(15090) 또는 인트라 예측부(15100)로부터 출력된 예측 신호에 더함으로써 복원(reconstructed) 신호(복원 픽처, 복원 블록, 복원 샘플 어레이)가 생성될 수 있다. 스킵 모드가 적용된 경우와 같이 처리 대상 블록에 대한 레지듀얼이 없는 경우, 예측된 블록이 복원 블록으로 사용될 수 있다. 가산부(155)는 복원부 또는 복원 블록 생성부라고 불릴 수 있다. 생성된 복원 신호는 현재 픽처 내 다음 처리 대상 블록의 인트라 예측을 위하여 사용될 수 있고, 후술하는 바와 같이 필터링을 거쳐서 다음 픽처의 인터 예측을 위하여 사용될 수도 있다.
필터링부(15070)는 복원 신호에 필터링을 적용하여 주관적/객관적 화질을 향상시킬 수 있다. 예를 들어 필터링부(15070)은 복원 픽처에 다양한 필터링 방법을 적용하여 수정된(modified) 복원 픽처를 생성할 수 있고, 수정된 복원 픽처를 메모리(15080), 구체적으로 메모리(15080)의 DPB에 저장할 수 있다. 다양한 필터링 방법은 예를 들어, 디블록킹 필터링, 샘플 적응적 오프셋(sample adaptive offset), 적응적 루프 필터(adaptive loop filter), 양방향 필터(bilateral filter) 등을 포함할 수 있다. 필터링부(15070)은 각 필터링 방법에 대한 설명에서 후술하는 바와 같이 필터링에 관한 다양한 정보를 생성하여 엔트로피 인코딩부(15110)로 전달할 수 있다. 필터링 관한 정보는 엔트로피 인코딩부(15110)에서 인코딩되어 비트스트림 형태로 출력될 수 있다.
메모리(15080)에 전송된 수정된 복원 픽처는 인터 예측부(15090)에서 참조 픽처로 사용될 수 있다. 인코딩 장치는 이를 통하여 인터 예측이 적용되는 경우, 인코딩 장치(15000)와 디코딩 장치에서의 예측 미스매치를 피할 수 있고, 부호화 효율도 향상시킬 수 있다.
메모리(15080) DPB는 수정된 복원 픽처를 인터 예측부(15090)에서의 참조 픽처로 사용하기 위해 저장할 수 있다. 메모리(15080)는 현재 픽처 내 움직임 정보가 도출된(또는 인코딩된) 블록의 움직임 정보 및/또는 이미 복원된 픽처 내 블록들의 움직임 정보를 저장할 수 있다. 저장된 움직임 정보는 공간적 주변 블록의 움직임 정보 또는 시간적 주변 블록의 움직임 정보로 활용하기 위하여 인터 예측부(15090)에 전달할 수 있다. 메모리(15080)는 현재 픽처 내 복원된 블록들의 복원 샘플들을 저장할 수 있고, 인트라 예측부(15100)에 전달할 수 있다.
한편, 상술한 예측, 변환, 양자화 절차 중 적어도 하나가 생략될 수도 있다. 예를 들어, PCM(pulse coding mode)가 적용되는 블록에 대하여는 예측, 변환, 양자화 절차를 생략하고 원본 샘플의 값이 그대로 인코딩되어 비트스트림으로 출력될 수도 있다.
도16은 실시예들에 따른 V-PCC 디코딩 프로세스(decoding process)의 예시를 나타낸다.
V-PCC 디코딩 프로세스 또는 V-PCC 디코더는 도4의 V-PCC 인코딩 프로세스(또는 인코더)의 역과정을 따를 수 있다. 도16의 각 구성요소는 소프트웨어, 하드웨어, 프로세서, 및/또는 그것들의 조합에 대응할 수 있다.
디멀티플렉서(demultiplexer, 16000)는 컴프레스된 비트스트림을 디멀티플렉싱하여 컴프로스된 텍스쳐 이미지, 컴프레스된 지오메트리 이미지, 컴프레스된 오큐판시 맵, 컴프레스된 어실러리 패치 인포메이션을 출력한다.
비디오 디컴프레션(video decompression, 16001, 16002) 또는 비디오 디컴프레서는 컴프레스된 텍스쳐 이미지 및 컴프레스된 지오메트리 이미지 각각을 디컴프레션(또는 디코딩)한다.
오큐판시 맵 디컴프레션(occupancy map decompression, 16003) 또는 오큐판시 맵 디컴프레서는 컴프레스된 오큐판시 맵을 디컴프레션한다.
어실러리 패치 인포 디컴프레션(auxiliary patch infor decompression, 16004) 또는 어실러리 패치 정보 디컴프레서는 어실러리 패치 정보를 디컴프레션한다.
지오메트리 리컨스럭션(geometry reconstruction, 16005) 또는 지오메트리 리컨스트럭터는 디컴프레스된 지오메트리 이미지, 디컴프레스된 어큐판시 맵, 및/또는 디컴프레스된 어실러리 패치 정보에 기반하여 지오메트리 정보를 복원(재구성)한다. 예를 들어, 인코딩과정에서 변경된 지오메트리를 리컨스럭션할 수 있다.
스무딩(smoothing, 16006) 또는 스무더는 재구성된 지오메트리에 대해 스무딩을 적용할 수 있다. 예를 들어, 스무딩 필터링이 적용될 수 있다.
텍스쳐 리컨스럭션(texture reconstruction, 16007) 또는 텍스쳐 리컨스트럭터는 디컴프레스된 텍스쳐 이미지 및/또는 스무딩된 지오메트리로부터 텍스쳐를 재구성한다.
컬러 스무딩(color smoothing, 16008) 또는 컬러 스무더는 재구성된 텍스쳐로부터 컬러 값을 스무딩한다. 예들 들어, 스무딩 필처링이 적용될 수 있다.
그 결과, 재구성된 포인트 클라우드 데이터가 생성될 수 있다.
도면은 압축된 occupancy map, geometry image, texture image, auxiliary path information 복호화하여 point cloud를 재구성하기 위한 V-PCC의 decoding process를 도시하여 보여주고 있다. 같다. 실시예들에 따른 각 프로세스의 동작은 다음과 같다.
비디오 디컴프레션(Video decompression, 16001, 16002)
앞서 설명한 video compression의 역과정으로, HEVC, VVC 등의 2D 비디오 코덱 을 이용하여, 앞서 설명한 과정으로 생성된 geometry image, texture image, occupancy map image 등의 compressed bitstream을 복호화하는 과정이다.
도17은 실시예들에 따른 2D 비디오/이미지 디코더(2D Video/Image Decoder)의 예시를 나타낸다.
2D 비디오/이미지 디코더는 도15의 2D 비디오/이미지 인코더의 역과정을 따를 수 있다.
도17의 2D 비디오/이미지 디코더는 도16의 비디오 디컴프레션(Video decompression) 또는 비디오 디컴프레서의 실시예로서, 비디오/영상 신호의 디코딩이 수행되는 2D 비디오/이미지 디코더(17000)의 개략적인 블록도를 나타낸다. 2D 비디오/이미지 디코더(17000)는 도1의 포인트 클라우드 비디오 디코더에 포함될 수 있고, 또는 내/외부 컴포넌트로 구성될 수도 있다. 도17의 각 구성요소는 소프트웨어, 하드웨어, 프로세서 및/또는 그것들의 조합에 대응할 수 있다.
여기서 입력 비트스트림은 상술한 geometry image, texture image (attribute(s) image), occupancy map image 등에 대한 비트스트림을 포함할 수 있다. 복원 영상(또는 출력 영상, 디코딩된 영상)은 상술한 geometry image, texture image (attribute(s) image), occupancy map image에 대한 복원 영상을 나타낼 수 있다.
도면을 참조하면, 인터 예측부(17070) 및 인트라 예측부(17080)를 합쳐서 예측부라고 불릴 수 있다. 즉, 예측부는 인터 예측부(180) 및 인트라 예측부(185)를 포함할 수 있다. 역양자화부(17020), 역변환부(17030)를 합쳐서 레지듀얼 처리부라고 불릴 수 있다. 즉, 레지듀얼 처리부는 역양자화부(17020), 역변환부(17030)을 포함할 수 있다. 상술한 엔트로피 디코딩부(17010), 역양자화부(17020), 역변환부(17030), 가산부(17040), 필터링부(17050), 인터 예측부(17070) 및 인트라 예측부(17080)는 실시예에 따라 하나의 하드웨어 컴포넌트(예를 들어 디코더 또는 프로세서)에 의하여 구성될 수 있다. 또한 메모리(170)는 DPB(decoded picture buffer)를 포함할 수 있고, 디지털 저장 매체에 의하여 구성될 수도 있다.
비디오/영상 정보를 포함하는 비트스트림이 입력되면, 디코딩 장치(17000)는 도 0.2-1의 인코딩 장치에서 비디오/영상 정보가 처리된 프로세스에 대응하여 영상을 복원할 수 있다. 예를 들어, 디코딩 장치(17000)는 인코딩 장치에서 적용된 처리 유닛을 이용하여 디코딩을 수행할 수 있다. 따라서 디코딩의 처리 유닛은 예를 들어 코딩 유닛일 수 있고, 코딩 유닛은 코딩 트리 유닛 또는 최대 코딩 유닛으로부터 쿼드 트리 구조 및/또는 바이너리 트리 구조를 따라서 분할될 수 있다. 그리고, 디코딩 장치(17000)를 통해 디코딩 및 출력된 복원 영상 신호는 재생 장치를 통해 재생될 수 있다.
디코딩 장치(17000)는 인코딩 장치로부터 출력된 신호를 비트스트림 형태로 수신할 수 있고, 수신된 신호는 엔트로피 디코딩부(17010)를 통해 디코딩될 수 있다. 예를 들어, 엔트로피 디코딩부(17010)는 비트스트림을 파싱하여 영상 복원(또는 픽처 복원)에 필요한 정보(ex. 비디오/영상 정보)를 도출할 수 있다. 예컨대, 엔트로피 디코딩부(17010)는 지수 골롬 부호화, CAVLC 또는 CABAC 등의 코딩 방법을 기초로 비트스트림 내 정보를 디코딩하고, 영상 복원에 필요한 신택스 엘리먼트의 값, 레지듀얼에 관한 변환 계수의 양자화된 값 들을 출력할 수 있다. 보다 상세하게, CABAC 엔트로피 디코딩 방법은, 비트스트림에서 각 구문 요소에 해당하는 빈을 수신하고, 디코딩 대상 구문 요소 정보와 주변 및 디코딩 대상 블록의 디코딩 정보 혹은 이전 단계에서 디코딩된 심볼/빈의 정보를 이용하여 문맥(context) 모델을 결정하고, 결정된 문맥 모델에 따라 빈(bin)의 발생 확률을 예측하여 빈의 산술 디코딩(arithmetic decoding)를 수행하여 각 구문 요소의 값에 해당하는 심볼을 생성할 수 있다. 이때, CABAC 엔트로피 디코딩 방법은 문맥 모델 결정 후 다음 심볼/빈의 문맥 모델을 위해 디코딩된 심볼/빈의 정보를 이용하여 문맥 모델을 업데이트할 수 있다. 엔트로피 디코딩부(17010)에서 디코딩된 정보 중 예측에 관한 정보는 예측부(인터 예측부(17070) 및 인트라 예측부(265))로 제공되고, 엔트로피 디코딩부(17010)에서 엔트로피 디코딩이 수행된 레지듀얼 값, 즉 양자화된 변환 계수들 및 관련 파라미터 정보는 역양자화부(17020)로 입력될 수 있다. 또한, 엔트로피 디코딩부(17010)에서 디코딩된 정보 중 필터링에 관한 정보는 필터링부(17050)으로 제공될 수 있다. 한편, 인코딩 장치로부터 출력된 신호를 수신하는 수신부(미도시)가 디코딩 장치(17000)의 내/외부 엘리먼트로서 더 구성될 수 있고, 또는 수신부는 엔트로피 디코딩부(17010)의 구성요소일 수도 있다.
역양자화부(17020)에서는 양자화된 변환 계수들을 역양자화하여 변환 계수들을 출력할 수 있다. 역양자화부(17020)는 양자화된 변환 계수들을 2차원의 블록 형태로 재정렬할 수 있다. 이 경우 재정렬은 인코딩 장치에서 수행된 계수 스캔 순서를 기반하여 재정렬을 수행할 수 있다. 역양자화부(17020)는 양자화 파라미터(예를 들어 양자화 스텝 사이즈 정보)를 이용하여 양자화된 변환 계수들에 대한 역양자화를 수행하고, 변환 계수들(transform coefficient)를 획득할 수 있다.
역변환부(17030)에서는 변환 계수들를 역변환하여 레지듀얼 신호(레지듀얼 블록, 레지듀얼 샘플 어레이)를 획득하게 된다.
예측부는 현재 블록에 대한 예측을 수행하고, 현재 블록에 대한 예측 샘플들을 포함하는 예측된 블록(predicted block)을 생성할 수 있다. 예측부는 엔트로피 디코딩부(17010)로부터 출력된 예측에 관한 정보를 기반으로 현재 블록에 인트라 예측이 적용되는지 또는 인터 예측이 적용되는지 결정할 수 있고, 구체적인 인트라/인터 예측 모드를 결정할 수 있다.
인트라 예측부(265)는 현재 픽처 내의 샘플들을 참조하여 현재 블록을 예측할 수 있다. 참조되는 샘플들은 예측 모드에 따라 현재 블록의 주변(neighbor)에 위치할 수 있고, 또는 떨어져서 위치할 수도 있다. 인트라 예측에서 예측 모드들은 복수의 비방향성 모드와 복수의 방향성 모드를 포함할 수 있다. 인트라 예측부(265)는 주변 블록에 적용된 예측 모드를 이용하여, 현재 블록에 적용되는 예측 모드를 결정할 수도 있다.
인터 예측부(17070)는 참조 픽처 상에서 움직임 벡터에 의해 특정되는 참조 블록(참조 샘플 어레이)을 기반으로, 현재 블록에 대한 예측된 블록을 유도할 수 있다. 이때, 인터 예측 모드에서 전송되는 움직임 정보의 양을 줄이기 위해 주변 블록과 현재 블록 간의 움직임 정보의 상관성에 기초하여 움직임 정보를 블록, 서브블록 또는 샘플 단위로 예측할 수 있다. 움직임 정보는 움직임 벡터 및 참조 픽처 인덱스를 포함할 수 있다. 움직임 정보는 인터 예측 방향(L0 예측, L1 예측, Bi 예측 등) 정보를 더 포함할 수 있다. 인터 예측의 경우에, 주변 블록은 현재 픽처 내에 존재하는 공간적 주변 블록(spatial neighboring block)과 참조 픽처에 존재하는 시간적 주변 블록(temporal neighboring block)을 포함할 수 있다. 예를 들어, 인터 예측부(17070)는 주변 블록들을 기반으로 움직임 정보 후보 리스트를 구성하고, 수신한 후보 선택 정보를 기반으로 현재 블록의 움직임 벡터 및/또는 참조 픽처 인덱스를 도출할 수 있다. 다양한 예측 모드를 기반으로 인터 예측이 수행될 수 있으며, 예측에 관한 정보는 현재 블록에 대한 인터 예측의 모드를 지시하는 정보를 포함할 수 있다.
가산부(17040)는 획득된 레지듀얼 신호를 인터 예측부(17070) 또는 인트라 예측부(265)로부터 출력된 예측 신호(예측된 블록, 예측 샘플 어레이)에 더함으로써 복원 신호(복원 픽처, 복원 블록, 복원 샘플 어레이)를 생성할 수 있다. 스킵 모드가 적용된 경우와 같이 처리 대상 블록에 대한 레지듀얼이 없는 경우, 예측된 블록이 복원 블록으로 사용될 수 있다.
가산부(17040)는 복원부 또는 복원 블록 생성부라고 불릴 수 있다. 생성된 복원 신호는 현재 픽처 내 다음 처리 대상 블록의 인트라 예측을 위하여 사용될 수 있고, 후술하는 바와 같이 필터링을 거쳐서 다음 픽처의 인터 예측을 위하여 사용될 수도 있다.
필터링부(17050)는 복원 신호에 필터링을 적용하여 주관적/객관적 화질을 향상시킬 수 있다. 예를 들어 필터링부(17050)는 복원 픽처에 다양한 필터링 방법을 적용하여 수정된(modified) 복원 픽처를 생성할 수 있고, 수정된 복원 픽처를 메모리(17060), 구체적으로 메모리(17060)의 DPB에 전송할 수 있다. 다양한 필터링 방법은 예를 들어, 디블록킹 필터링, 샘플 적응적 오프셋(sample adaptive offset), 적응적 루프 필터(adaptive loop filter), 양방향 필터(bilateral filter) 등을 포함할 수 있다.
메모리(17060)의 DPB에 저장된 (수정된) 복원 픽처는 인터 예측부(17070)에서 참조 픽쳐로 사용될 수 있다. 메모리(17060)는 현재 픽처 내 움직임 정보가 도출된(또는 디코딩된) 블록의 움직임 정보 및/또는 이미 복원된 픽처 내 블록들의 움직임 정보를 저장할 수 있다. 저장된 움직임 정보는 공간적 주변 블록의 움직임 정보 또는 시간적 주변 블록의 움직임 정보로 활용하기 위하여 인터 예측부(17070)에 전달할 수 있다. 메모리(170)는 현재 픽처 내 복원된 블록들의 복원 샘플들을 저장할 수 있고, 인트라 예측부(17080)에 전달할 수 있다.
본 명세서에서, 인코딩 장치(100)의 필터링부(160), 인터 예측부(180) 및 인트라 예측부(185)에서 설명된 실시예들은 각각 디코딩 장치(17000)의 필터링부(17050), 인터 예측부(17070) 및 인트라 예측부(17080)에도 동일 또는 대응되도록 적용될 수 있다.
한편, 상술한 예측, 변환, 양자화 절차 중 적어도 하나가 생략될 수도 있다. 예를 들어, PCM(pulse coding mode)가 적용되는 블록에 대하여는 예측, 변환, 양자화 절차를 생략하고 디코딩된 샘플의 값이 그대로 복원 영상의 샘플로 사용될 수도 있다.
오큐판시 맵 디컴프레션(Occupancy map decompression, 16003)
앞서 설명한 occupancy map compression의 역과정으로, 압축된 occupancy map bitstream을 복호화하여 occupancy map을 복원하기 위한 과정이다.
어실러리 패치 인포 디컴프레션(Auxiliary patch info decompression, 16004)
앞서 설명한 auxiliary patch info compression의 역과정을 수행하고, 압축된 auxiliary patch info bitstream 를 복호화하여 auxiliary patch info를 복원할 수 있다.
지오메트리 리컨스럭션(Geometry reconstruction, 16005)
앞서 설명한 geometry image generation의 역과정이다. 먼저, 복원된 occupancy map 과 auxiliary patch info에 포함되는 patch의 2D 위치/크기 정보 및 block과 patch의 맵핑 정보를 이용하여 geometry image에서 patch를 추출한다. 이후 추출된 patch의 geometry image와 auxiliary patch info에 포함되는 patch의 3D 위치 정보를 이용하여 point cloud를 3차원 공간상에 복원한다. 하나의 patch내에 존재하는 임의의 점 (u, v)에 해당하는 geometry 값을 g(u, v)라 하고, patch의 3차원 공간상 위치의 normal 축, tangent 축, bitangent 축 좌표값을 (d0, s0, r0)라 할 때, 점 (u, v)에 맵핑되는 3차원 공간상 위치의 normal 축, tangent 축, bitangent 축 좌표값인 d(u, v), s(u, v), r(u, v)는 다음과 같이 나타낼 수 있다.
d(u, v) = d0 + g(u, v)
s(u, v) = s0 + u
r(u, v) = r0 + v
스무딩(Smoothing, 16006)
앞서 설명한 encoding process에서의 smoothing과 동일하며, 압축 과정에서 발생하는 화질의 열화로 인해 patch 경계면에서 발생할 수 있는 불연속성을 제거하기 위한 과정이다.
텍스쳐 리컨스럭션(Texture reconstruction, 16007)
Smoothing된 point cloud를 구성하는 각 점들에 color값을 부여하여 color point cloud를 복원하는 과정이다. 2.4에서 설명한 geometry reconstruction 과정에서의 geometry image와 point cloud의 맵핑 정보를 이용하여 2D 공간에서 geometry image에서와 동일한 위치의 texture image 픽셀에 해당되는 color 값들을, 3D 공간에서 동일한 위치에 대응되는 point cloud의 점에 부여함으로써 수행될 수 있다.
컬러 스무딩(Color smoothing, 16008)
앞서 설명한 geometry smoothing의 과정과 유사하며, 압축 과정에서 발생하는 화질의 열화로 인해 patch 경계면에서 발생할 수 있는 color 값들의 불연속성을 제거하기 위한 작업이다. 다음과 같은 과정으로 수행될 수 있다.
① K-D tree 등을 이용하여 복원된 color point cloud를 구성하는 각 점들의 인접점들을 산출한다. 2.5절에서 설명한 geometry smoothing 과정에서 산출된 인접점 정보를 그대로 이용할 수도 있다.
② 각 점들에 대하여, 해당 점이 patch 경계면에 위치하는지를 판단한다. 2.5절에서 설명한 geometry smoothing 과정에서 산출된 경계면 정보를 그대로 이용할 수도 있다.
③ 경계면에 존재하는 점의 인접점들에 대하여, color 값의 분포를 조사하여 smoothing 여부를 판단한다. 일례로, 휘도값의 entropy가 경계 값 (threshold local entry) 이하일 경우 (유사한 휘도 값들이 많을 경우), edge가 아닌 부분으로 판단하여 smoothing을 수행할 수 있다. Smoothing의 방법으로 인접접들의 평균값으로 해당 점의 color값을 바꾸는 방법 등이 사용될 수 있다.
도18은 실시예들에 따른 송신 장치의 동작 흐름도의 예시를 나타낸다.
실시예들의 따른 송신 장치는 도1의 송신 장치, 도4의 인코딩 프로세스, 도15의 2D 비디오/이미지 인코더에 대응하거나 그것들의 동작을 일부/전부 수행할 수 있다. 송신 장치의 각 구성요소는 소프트웨어, 하드웨어, 프로세서 및/또는 그것들의 조합에 대응할 수 있다.
V-PCC를 이용한 포인트 클라우드 데이터의 압축 및 전송을 위한 송신단의 동작 과정은 도면과 같은 수 있다.
실시예들에 따른 포인트 클라우드 데이터 송신 장치는 송신 장치 등으로 지칭될 수 있다.
패치 생성부(18000) 관련하여, 먼저, 포인트 클라우드(point cloud)의 2D 이미지 맵핑을 위한 패치 (patch)를 생성한다. 패치 생성의 결과물로 부가 패치 정보가 생성되며, 해당 정보는 지오메트리 이미지 (geometry image) 생성, 텍스처 이미지 (texture image) 생성, 스무딩 (smoothing)을 위한 지오메트리 복원과정에 사용될 수 있다.
패치 패킹부(18001) 관련하여, 생성된 패치들은 2D 이미지 안에 맵핑하는 패치 패킹 과정을 거치게 된다. 패치 패킹의 결과물로 오큐판시 맵 (occupancy map)을 생성할 수 있으며, 오큐판시 맵은 지오메트리 이미지 생성, 텍스처 이미지 생성, 스무딩을 위한 지오메트리 복원과정에 사용될 수 있다.
지오메트리 이미지 생성부(18002)는 부가 패치 정보와 오큐판시 맵을 이용하여 지오메트리 이미지를 생성하며, 생성된 지오메트리 이미지는 비디오 부호화를 통해 하나의 비트스트림 (bitstream)으로 부호화된다.
부호화 전처리(18003)는 이미지 패딩 절차를 포함할 수 있다. 생성된 지오메트리 이미지 또는 부호화된 지오메트리 비트스트림을 복호화하여 재생성된 지오메트리 이미지는 3차원 지오메트리 복원에 사용될 수 있고 이후 스무딩 과정을 거칠 수 있다.
텍스처 이미지 생성부(18004)는 (스무딩된) 3차원 지오메트리와 포인트 클라우드, 부가 패치 정보 및 오큐판시 맵을 이용하여 텍스처 이미지를 생성할 수 있다. 생성된 텍스처 이미지는 하나의 비디오 비트스트림으로 부호화될 수 있다.
메타데이터 부호화부(18005)는 부가 패치 정보를 하나의 메타데이터 비트스트림으로 부호화할 수 있다.
비디오 부호화부(18006)는 오큐판시 맵을 하나의 비디오 비트스트림으로 부호화할 수 있다.
다중화부(18007)는 생성된 지오메트리, 텍스처 이미지, 오큐판시 맵의 비디오 비트스트림과 부가 패치 정보 메타데이터 비트스트림은 하나의 비트스트림으로 다중화한다.
송신부(18008)는 비트스트림을 수신단에 전송될 수 있다. 또는 생성된 지오메트리, 텍스처 이미지, 오큐판시 맵의 비디오 비트스트림과 부가 패치 정보 메타데이터 비트스트림은 하나 이상의 트랙 데이터로 파일이 생성되거나 세그먼트로 인캡슐레이션 되어 송신부를 통해 수신단에 전송 될 수 있다.
도19는 실시예들에 따른 수신 장치의 동작 흐름도의 예시를 나타낸다.
실시예들에 따른 수신 장치는 도1의 수신 장치, 도16의 디코딩 프로세스, 도17의 2D 비디오/이미지 인코더에 대응하거나 그것들의 동작을 일부/전부 수행할 수 있다. 수신 장치의 각 구성요소는 소프트웨어, 하드웨어, 프로세서 및/또는 그것들의 조합에 대응할 수 있다.
V-PCC를 이용한 포인트 클라우드 데이터의 수신 및 복원을 위한 수신단의 동작 과정은 도면과 같은 수 있다. V-PCC 수신단의 동작은 도18의 V-PCC 송신단의 동작의 역과정을 따를 수 있다.
실시예들에 따른 포인트 클라우드 데이터 수신 장치는 수신 장치 등으로 지칭될 수 있다.
수신된 포인트 클라우드의 비트스트림은 파일/세그먼트 디캡슐레이션 후 압축된 지오메트리 이미지, 텍스처 이미지, 오큐판시 맵의 비디오 비트스트림들과 부가 패치 정보 메테데이터 비트스트림으로 역다중화부(19000)에 의해 역다중화된다. 비디오 복호화부(19001)와 메타데이터 복호화부(19002)는 역다중화된 비디오 비트스트림들과 메타데이터 비트스트림을 복호화한다. 지오메트리 복원부(19003)에 의해 복호화된 지오메트리 이미지와 오큐판시 맵 및 부가 패치 정보를 이용하여 3차원 지오메트리가 복원되며 이후 스무더(19004)에 의한 스무딩 과정을 거친다. 스무딩된 3차원 지오메트리에 텍스처 이미지를 이용하여 컬러값을 부여함으로써 컬러 포인트 클라우드 영상/픽처가 텍스쳐 복원부(19005)에 의해 복원될 수 있다. 이후 객관적/주관적 비주얼 퀄리티 향상을 위하여 컬러 스무딩 (color smoothing)과정을 추가적으로 수행할 수 있으며, 이를 통하여 도출된 수정된(modified) 포인트 클라우드 영상/픽처는 렌더링 과정을 통하여(ex. by 포인트 클라우드 렌더러)를 통해 사용자에게 보여진다. 한편, 컬러 스무딩 과정은 경우에 따라 생략될 수 있다.
도20은 실시예들에 따른 포인트 클라우드 데이터 송수신 방법/장치와 연동 가능한 구조의 예시를 나타낸다.
실시예들에 따른 구조는 서버(2360), 로봇(2010), 자율 주행 차량(2020), XR 장치(2030), 스마트폰(2040), 가전(2050) 및/또는 HMD(2070) 중에서 적어도 하나 이상이 클라우드 네트워크(2010)와 연결된다. 여기서, 로봇(2010), 자율 주행 차량(2020), XR 장치(2030), 스마트폰(2040) 또는 가전(2050) 등을 장치라 칭할 수 있다. 또한, XR 장치(2030)는 실시예들에 따른 포인트 클라우드 데이터 (PCC) 장치에 대응되거나 PCC장치와 연동될 수 있다.
클라우드 네트워크(2000)는 클라우드 컴퓨팅 인프라의 일부를 구성하거나 클라우드 컴퓨팅 인프라 안에 존재하는 네트워크를 의미할 수 있다. 여기서, 클라우드 네트워크(2000)는 3G 네트워크, 4G 또는 LTE(Long Term Evolution) 네트워크 또는 5G 네트워크 등을 이용하여 구성될 수 있다.
서버(2360)는 로봇(2010), 자율 주행 차량(2020), XR 장치(2030), 스마트폰(2040), 가전(2050) 및/또는 HMD(2070) 중에서 적어도 하나 이상과 클라우드 네트워크(2000)을 통하여 연결되고, 연결된 장치들(2010 내지 2070)의 프로세싱을 적어도 일부를 도울 수 있다.
HMD (Head-Mount Display)(2070)는 실시예들에 따른 XR 디바이스 및/또는 PCC 디바이스가 구현될 수 있는 타입 중 하나를 나타낸다. 실시예들에 따른HMD 타입의 디바이스는, 커뮤니케이션 유닛, 컨트롤 유닛, 메모리 유닛, I/O 유닛, 센서 유닛, 그리고 파워 공급 유닛 등을 포함한다.
이하에서는, 상술한 기술이 적용되는 장치(2010 내지 2070)의 다양한 실시 예들을 설명한다. 여기서, 도20에 도시된 장치(2000 내지 2700)는 상술한 실시예들에 따른 포인트 클라우드 데이터 송수신 장치와 연동/결합될 수 있다.
<PCC+XR> XR/PCC 장치(2030)는 PCC 및/또는 XR(AR+VR) 기술이 적용되어, HMD(Head-Mount Display), 차량에 구비된 HUD(Head-Up Display), 텔레비전, 휴대폰, 스마트 폰, 컴퓨터, 웨어러블 디바이스, 가전 기기, 디지털 사이니지, 차량, 고정형 로봇이나 이동형 로봇 등으로 구현될 수도 있다.
XR/PCC 장치(2030)는 다양한 센서들을 통해 또는 외부 장치로부터 획득한 3차원 포인트 클라우드 데이터 또는 이미지 데이터를 분석하여 3차원 포인트들에 대한 위치 데이터 및 속성 데이터를 생성함으로써 주변 공간 또는 현실 객체에 대한 정보를 획득하고, 출력할 XR 객체를 렌더링하여 출력할 수 있다. 예컨대, XR/PCC 장치(2030)는 인식된 물체에 대한 추가 정보를 포함하는 XR 객체를 해당 인식된 물체에 대응시켜 출력할 수 있다.
<PCC+자율주행+XR> 자율 주행 차량(2020)은 PCC 기술 및 XR 기술이 적용되어, 이동형 로봇, 차량, 무인 비행체 등으로 구현될 수 있다.
XR/PCC 기술이 적용된 자율 주행 차량(2020)은 XR 영상을 제공하는 수단을 구비한 자율 주행 차량이나, XR 영상 내에서의 제어/상호작용의 대상이 되는 자율 주행 차량 등을 의미할 수 있다. 특히, XR 영상 내에서의 제어/상호작용의 대상이 되는 자율 주행 차량(2020)은 XR 장치(2030)와 구분되며 서로 연동될 수 있다.
XR/PCC영상을 제공하는 수단을 구비한 자율 주행 차량(2020)은 카메라를 포함하는 센서들로부터 센서 정보를 획득하고, 획득한 센서 정보에 기초하여 생성된 XR/PCC 영상을 출력할 수 있다. 예컨대, 자율 주행 차량은 HUD를 구비하여 XR/PCC 영상을 출력함으로써, 탑승자에게 현실 객체 또는 화면 속의 객체에 대응되는 XR/PCC 객체를 제공할 수 있다.
이때, XR/PCC 객체가 HUD에 출력되는 경우에는 XR/PCC 객체의 적어도 일부가 탑승자의 시선이 향하는 실제 객체에 오버랩되도록 출력될 수 있다. 반면, XR/PCC 객체가 자율 주행 차량의 내부에 구비되는 디스플레이에 출력되는 경우에는 XR/PCC 객체의 적어도 일부가 화면 속의 객체에 오버랩되도록 출력될 수 있다. 예컨대, 자율 주행 차량은 차로, 타 차량, 신호등, 교통 표지판, 이륜차, 보행자, 건물 등과 같은 객체와 대응되는 XR/PCC 객체들을 출력할 수 있다.
실시예들에 의한 VR (Virtual Reality) 기술, AR (Augmented Reality) 기술, MR (Mixed Reality) 기술 및/또는 PCC(Point Cloud Compression)기술은, 다양한 디바이스에 적용 가능하다.
즉, VR 기술은, 현실 세계의 객체나 배경 등을 CG 영상으로만 제공하는 디스플레이 기술이다. 반면, AR 기술은, 실제 사물 영상 위에 가상으로 만들어진 CG 영상을 함께 보여 주는 기술을 의미한다. 나아가, MR 기술은, 현실세계에 가상 객체들을 섞고 결합시켜서 보여준다는 점에서 전술한 AR 기술과 유사하다. 그러나, AR 기술에서는 현실 객체와 CG 영상으로 만들어진 가상 객체의 구별이 뚜렷하고, 현실 객체를 보완하는 형태로 가상 객체를 사용하는 반면, MR 기술에서는 가상 객체가 현실 객체와 동등한 성격으로 간주된다는 점에서 AR 기술과는 구별이 된다. 보다 구체적으로 예를 들면, 전술한 MR 기술이 적용된 것이 홀로그램 서비스 이다.
다만, 최근에는 VR, AR, MR 기술을 명확히 구별하기 보다는 XR (extended Reality) 기술로 부르기도 한다. 따라서, 본 발명의 실시예들은 VR, AR, MR, XR 기술 모두에 적용 가능하다. 이런 한 기술은 PCC, V-PCC, G-PCC 기술 기반 인코딩/디코딩이 적용될 수 있다.
실시예들에 따른 PCC방법/장치는 자율 주행 서비스를 제공하는 차량에 적용될 수 있다.
자율 주행 서비스를 제공하는 차량은 PCC 디바이스와 유/무선 통신이 가능하도록 연결된다.
실시예들에 따른 포인트 클라우드 데이터 (PCC) 송수신 장치는 차량과 유/무선 통신이 가능하도록 연결된 경우, 자율 주행 서비스와 함께 제공할 수 있는 AR/VR/PCC 서비스 관련 콘텐트 데이터를 수신/처리하여 차량에 전송할 수 있다. 또한 포인트 클라우드 데이터 송수신 장치 차량에 탑재된 경우, 포인트 클라우드 송수신 장치는 사용자 인터페이스 장치를 통해 입력된 사용자 입력 신호에 따라 AR/VR/PCC 서비스 관련 콘텐트 데이터를 수신/처리하여 사용자에게 제공할 수 있다. 실시예들에 따른 차량 또는 사용자 인터페이스 장치는 사용자 입력 신호를 수신할 수 있다. 실시예들에 따른 사용자 입력 신호는 자율 주행 서비스를 지시하는 신호를 포함할 수 있다.
실시예들에 따른 포인트 클라우드 데이터 송신 방법/장치는 도1 송신 장치(10000), 포인트 클라우드 비디오 인코더(10002), 도4 인코딩 프로세스, 도15 비디오/이미지 인코더, 도18 송신 장치, 도20 XR디바이스(1730), 도40 송신 장치 등을 지칭할 수 있다. 송신 방법/장치의 각 구성요소는 하드웨어, 소프트웨어, 메모리와 연결된 프로세서, 및/또는 그것들의 조합에 대응할 수 있다.
실시예들에 따른 포인트 클라우드 데이터 수신 방법/장치는 도2 수신 장치(10005), 포인트 클라우드 비디오 디코더(10008), 도16 디코딩 프로세스, 도17 비디오/이미지 디코더, 도19 수신 장치, 도20 XR디바이스(1730) 등을 지칭할 수 있다. 수신 방법/장치의 각 구성요소는 하드웨어, 소프트웨어, 메모리와 연결된 프로세서, 및/또는 그것들의 조합에 대응할 수 있다.
실시예들에 따른 포인트 클라우드 데이터 송수신 방법/장치는 실시예들에 따른 방법/장치로 줄여서 지칭할 수 있다.
도21은 실시예들에 따른 복셀화된 서페이스 라이트 필드 시퀀스를 나타낸다.
실시예들에 따른 방법/장치는 포인트 클라우드 데이터를 포함하는 서페이스 라이트 필드(Surface Light Field (SLF)) 데이터 셋을 압축하고, 복원할 수 있다.
실시예들에 따른 방법/장치는 V-PCC 기반 효과적인 Surface Light Field (SLF) 시퀀스 압축/복원을 위해서, 패치(Patch) 기반 카메라뷰 선택 및 어트리뷰트(Attribute) 비디오 생성 방법을 이용할 수 있다.
실시예들은 2D 비디오 코덱(Video codec)을 이용하여 3차원 포인트 클라우드(Point cloud) 데이터를 압축하는 방법인 Video-based Point Cloud Compression(V-PCC) 에 관한 것이다. 특히, V-PCC를 이용하여 일반적인 포인트 클라우드 데이터 셋 보다 많은 어트리뷰트(Attribute) 속성을 갖는 Surface Light Field (SLF) 데이터 셋을 압축할 때, 다른 방식에서 발생하는 낮은 압축률 및 높은 복잡도, 많은 코덱 인스턴스 사용 문제를 해결할 수 있다.
SLF 시퀀스를 V-PCC 로 압축하는 경우, 다른 방식보다 효과적인 압축 성능을 낼 수 있는 방법을 제공한다. 실시예들은 V-PCC 내부에서 SLF 시퀀스 데이터를 구성하는 카메라 뷰포인트들 중 유의미한 카메라 뷰포인트 선택 방법, Attribute 비디오 스트림 (Video stream) 생성 방법, 카메라뷰 선택 방법과 비디오 스트림 생성 방식에 대한 시그널링 (Signaling), 송신기와 수신에서의 비디오 스트림 처리 방법 등을 제공한다.
V-PCC 방법은 입력된 3차원 포인트 클라우드 데이터를 압축하기 위해, 입력된 데이터로부터 3개의 비디오 스트림, 1) 어큐판시 맵(Occupancy map) 비디오, 2) 지오메트리(Geometry) 비디오, 3) 어트리뷰트(Attribute) 비디오 스트림을 생성한다. 그리고 이들은 V-PCC 인코더 (Encoder)에서 2D 비디오 코덱을 이용하여 각각 압축된다. 이 중에서 어트리뷰트(Attribute) 비디오는 3D 포인트의 컬러 속성을 2D 이미지 상에 담기 위해 생성되는 것으로, Attribute 비디오가 텍스쳐(Texture) 타입의 정보를 갖는다면 이미지 상의 각 픽셀 값은 그에 상응하는 3D 포인트의 (R, G, B) 컬러 속성을 나타낸다.
한편, 복셀호된 서페이스 라이트 필드(Voxelized Surface Light Field) (VSLF) 데이터 셋은 실, 가상현실 등의 어플리케이션 상에서 3D 씭(Scene)을 렌더링할 때 보다 사실적인 표현과 자유로운 시점 이동으로 더욱 실감있는 경험을 사용자에게 제공하기 위해 만들어진 MPEG의 표준화 용 테스트 데이터 셋이다 (도21 참조). 이 SLF 데이터 셋은 일반적인 포인트 클라우드 데이터 셋이 하나의 포인트를 표현하는 컬러 속성으로 1개의 (R, G, B) 정보를 갖는 것과 달리, 서로 다른 카메라 뷰포인트의 개수만큼 (R, G, B) 정보를 가지고 있다. 그러므로 V-PCC 를 이용하여 일반적인 포인트 클라우드 데이터 셋을 압축하는 경우에는 하나의 Attribute 비디오를 생성하여 모든 포인트들의 각 컬러 속성을 표현, 압축할 수 있지만, SLF 데이터 셋을 압축하는 경우에는 하나의 Attribute 비디오로는 모든 컬러 속성을 담을 수 없기에 이런 특성을 고려한 추가적인 V-PCC 압축기술의 적용이 필요하다.
실시예들에 따른 V-PCC기법을 이용하여 SLF 시퀀스를 압축하기 위한 기술들이 있다.
실시예들은 V-PCC의 압축 과정에서 포인트 클라우드의 컬러 속성을 하나의 Attribute 비디오로 생성하는 방법을 모든 SLF 데이터 셋의 컬러 속성에 동일하게 적용하여 여러 개의 Attribute 비디오를 생성하고 압축할 수 있다.
도21의 시퀀스를 압축하는 경우에 대해, 실시예들은 1개의 Geometry 비디오, 1개의 Occupancy map, Auxiliary 정보와 더불어 13개의 Attribute 비디오를 생성한다. 그리고 V-PCC 에서 각각의 비디오 스트림을 2D 비디오 코덱을 이용하여 압축하는 것과 동일하게 13개의 Attribute 비디오 각각을 압축하도록 한다. 그 결과 동일 시퀀스를 SLF 데이터가 아닌 일반 포인트 클라우드 데이터 형태로 압축했을 때 (2 프레임, 최고화질 압축 조건)의 압축 데이터는 991099 Bytes 내지 0.9 Bytes 였고, SLF 데이터 셋 형태로 압축했을 때의 압축 데이터는 11083828 Bytes 내지 11 Mega bytes 의 용량을 가질 수 있다. SLF 시퀀스의 오리지널 데이터 사이즈가 크기 때문에 압축한 결과의 데이터 사이즈 역시 매우 큰 것으로 나타난다. 이는 실제 어플리케이션 상에 적용될 경우 높은 수준의 메모리 용량이 요구될 것으로 예측할 수 있다. 뿐만 아니라 13개의 Attribute 비디오를 각각 인코딩, 디코딩 하기 위해 많은 수의 비디오 코덱 인스턴스가 필요한 단점도 있다.
도22는 실시예들에 따른 2D 포인트 클라우드 데이터를 포인트 별로 액티비티 리젼(activity region ) 정보에 따라 구분하여 표시한 예시를 나타낸다.
실시예들은 이러한 과도한 메모리 사용 문제, 다른 표현으로는 낮은 압축률 문제를 해결하기 위해 보다 최적화된 기술을 포함할 수 있다. 모든 카메라 뷰포인트를 대표할 수 있는 대표 카메라를 선택하고, 이들의 카메라 뷰가 표현하는 컬러 정보만을 압축하도록 한다.
이를 위해 SLF 시퀀스의 모든 포인트들을 로우(low, 2200), 미디엄(medium, 2201), 하이 액티비티 리젼(high activity region, 2202)으로 분류할 수 있다 (도22 참고).
실시예들에 따른 방법/장치는 활동 영역(activity region) 정보를 함께 고려한 패치 세그멘티이션(patch segmentation)을 수행한다. 이 과정을 통해 하나의 패치(patch)에 속한 모든 포인트들은 동일한 활동 영역(activity region)을 갖게 되며, 하이 활동 영역(high activity region)을 갖는 패치에 대해서는 5개의 대표 카메라를 선택하고, 미디엄 활동 영역(medium activity region)을 갖는 패치에 대해서는 3개의 대표 카메라를 선택하고, 로우 활동 영역(low activity region)을 갖는 패치에 대해서는 1개의 대표 카메라를 선택할 수 있다. 이렇게 선택된 카메라가 표현하는 컬러 정보를 이용하여 attribute 비디오를 생성한다.
도23은 실시예들에 따른 SLF 데이터 셋으로부터 생성된 5개 어트리뷰트 비디오 및 1개 지오메트리 비디오 구성을 나타낸다.
최종적으로, 상술한 바와 같이 입력된 SLF 시퀀스에 대해 생성되는 Attribute 비디오는 총 5개이며 이 각각을 2D 비디오 코덱을 이용하여 압축하도록 한다 (도23 참고).
상술한 실시예는 40%의 압축률을 향상시키는 결과를 얻었지만, 복원된 뷰의 PSNR 성능이 크게 떨어지는 문제로 인해 추가적으로 더 개선할 수 있다.
실시예들에 따른 방법/장치는 설명한 방법의 메모리 요구량 문제를 해결하고, [높은 화질의 압축 성능을 제공하기 위한 추가 동작들을 더 제공할 수 있다.
구체적으로, V-PCC를 이용하여, SLF 시퀀스의 컬러 데이터를 효과적으로 압축하는 방법, 각 포인트에 대해 모든 카메라 뷰포인트의 컬러 정보를 압축하지 않고, 유의미한 카메라 뷰포인트의 컬러 정보만을 패치 단위로 선택하는 방법, 선택된 카메라 뷰포인트들의 컬러 정보를 이용하여 비디오를 생성, 압축하는 방법, V-PCC에서 사용되는 Attribute 비디오 개수를 최소화함으로써 인코딩, 디코딩 복잡도 및 필요한 코덱 인스턴스 개수를 최소화하는 방법, 송신단(혹은 V-PCC 인코더)과 수신단(혹은 V-PCC 디코더)에서 제안하는 SLF 시퀀스 압축 방식이 적용되었음을 알게 하는 시그널링 방법 등을 설명한다.
실시예들에 따른 용어인 어트리뷰트(Attribute) 비디오는 V-PCC 에서 지원하는 어틜뷰트(Attribute) 데이터(또는 어트리뷰트)를 이용하여 생성된 비디오를 의미하며, 만약 해당 Attribute 비디오가 Texture 데이터 타입을 갖는 경우에는 이를 텍스쳐(Texture) 비디오라고 지칭할 수 있다.
실시예들에 따른 SLF 시퀀스는 하나의 포인트에 대해 여러 대의 카메라로부터 획득된 컬러 정보로 구성된 데이터를 의미한다. 즉, SLF 시퀀스는 오브젝트를 표현하는 포인트 클라우드의 각 포인트에 대해 여러 개의 컬러 정보를 데이터로 포함하고 있다. 하나의 포인트에 대한 여러 개의 컬러 정보는 카메라의 촬영 위치, 각도 등에 따라 서로 다른 값을 가질 수 있다. 실시예들은SLF 시퀀스의 컬러 정보에 대해 Texture 비디오를 생성할 때, 패치 단위로 유의미한 컬러 정보들만 선택하고, 이들을 이용하여 비디오를 생성하는 방법을 제공한다.
먼저 패치 단위로 유의미한 컬러 정보를 선택하기 위해서는, SLF 데이터를 구성하는 전체 카메라 뷰포인트들 중에서 각 패치에 가장 유의미한 컬러 정보를 제공해주는 카메라 뷰포인트를 선택해야 한다. 실시예들은 패치 별 카메라 뷰포인트 선택을 위해 두 가지 방법을 이용한다.
제안하는 방법을 설명하기 위해 SLF 데이터를 구성하는 전체 카메라 뷰포인트 개수를 N, 패치 별 선택되는 카메라 뷰포인트의 개수를 M으로 정의한다.
도24는 실시예들에 따른 패치 별 카메라 뷰포인트 선택 방법 및 텍스쳐 비디오 생성하는 예시를 나타낸다.
실시예들에 따른 패치 별 카메라 뷰포인트 선택 방법 1(도24)
실시예들에 따른 포인트 클라우드 데이터 송신 방법/장치의 패치 패킹 단계 및 패치 패커(프로세서)가 이 동작을 수행할 수 있고, 실시예들에 따른 포인트 클라우드 데이터 수신 방법/장치의 패치 디코딩 과정에서 이 동작에 기반하여 패치를 디코딩할 수 있다.
실시예들은 SLF 데이터를 구성하는 N개의 카메라 뷰포인트 중에서, 패치가 향하는 방향과 가장 가깝게 마주하고 있는 M개의 카메라 뷰포인트를 선택하도록 한다 (도24 참고).
도24는 N=8, M=3인 예시를 나타낸다. N이 8이므로, 카메라가 0번부터 7번까지 있고, M이3이므로 3개의 카메라 뷰인트를 선택할 수 있다.
포인트 클라우드 렌더링 시 사용자가 실질적으로 볼 수 있는 포인트 영역은 각 영역의 전면부에 해당하는 경우가 대부분일 것이다. 그러므로 하나의 포인트에 대해 N개의 컬러 정보를 갖고 있다고 하여도 실제 사용자가 볼 수 있고 필요로 하는 정보는 일부 각도에 해당되는 컬러 정보로 한정될 가능성이 높다. 실시예들은 이러한 특성을 바탕으로 패치의 전면부 방향에 존재하는 카메라 뷰포인트로부터 획득된 컬러 정보가 가장 유효하고 정확하다고 판단하여, 해당 정보만을 선택하여 압축, 전송할 수 있도록 한다.
SLF 데이터 생성에 사용되는 N개의 카메라 뷰포인트는 각각 위치와 방향에 대한 카메라 매트릭스(camera matrix) 정보를 갖고 있다. 또한 포인트 클라우드 세그멘테이션(segmentation)을 통해 각 패치가 생성될 때 해당 패치가 프로젝션되는 방향이 결정되기 때문에 패치가 3차원 공간 상에서 향하는 방향을 노멀 벡터(normal vector)를 통해 알 수 있다. 그러므로 카메라 matrix 정보와 패치의 normal vector 간 연산을 통해, 패치와 가장 가깝게 마주하고 있는 카메라 뷰포인트를 결정할 수 있다. 카메라 matrix와 패치의 normal vector 간 연산은 벡터 내적(u
Figure PCTKR2021011600-appb-img-000005
v)이 되며, 이때 패치와 카메라 뷰포인트가 서로 마주하는 방향으로 가까울수록 벡터 내적 연산의 결과는 작은 값을 갖는다. 그러므로 모든 카메라 뷰포인트들의 카메라 matrix와 패치 normal vector 간 벡터 내적 연산을 수행하고, 그 중 가작 작은 값을 갖는 M개의 카메라 뷰포인트를 선택하면, 패치와 가장 가까이에서 마주보고 있는 카메라 뷰포인트들을 선택할 수 있다.
여기에서 M 값은 패치마다 다르게 설정할 수 있으며, 이 경우 카메라 matrix의 방향 벡터와 패치 normal vector가 이루는 각도가 일정 각도 범위를 벗어나는 경우는 제외하도록 추가적인 제약(constraint)을 줄 수 있다. 또는 타일 내에서는 동일한 M 값을 가지도록 할 수도 있으며, 또는 프레임 내에서는 동일한 M 값을 가지도록 할 수도 있다. 마찬가지로 동일 시퀀스에 대해서는 모두 같은 M 값을 갖도록 설정할 수도 있다. M 값의 설정에 대한 정보는 atlas 정보로 시그널링한다. 그리고 M개 만큼 선택된 카메라 뷰포인트의 정보는 각 패치마다 시그널링하여 전송한다. M의 값이 클수록 생성되는 Texture 비디오의 개수는 증가하지만 패치에 대한 뷰 커버리지 범위(View coverage range)는 넓어질 수 있다.
텍스쳐 비디오는 0번, 1번, 2번 등이 있을 수 있다. 각 텍스쳐 비디오는 텍스쳐 비디오에 대응하는 패치(들)을 포함하고, 패치들은 패치와 가장 가까운 카메라(들)로부터 획득되어 텍스쳐 비디오 포함될 수 있다.
예를 들어, 텍스쳐 비디오#0은 카메라#4, 카메라#6, 카메라#1로부터 획득된 패치들을 포함할 수 있다. 즉, 텍스쳐 비디오#0은 카메라#5, 카메라#4, 카메라#3 중에서 제일 가까운 카메라#4 기반 패치를 포함한다. 이와 같이, 오브젝트에 대한 포인트 클라우드 데이터를 상세하게 구성할 수 있고, 데이터 양이 증가하더라도 효율적으로 압축할 수 있다.
실시예들에 따른 패치 별 카메라 뷰포인트 선택 방법 2
실시예들에 따른 방법/장치는 패치 내 카메라 간 중복되는 어트리뷰트(컬러)가 있으면 이를 제외할 수 있다.
각 포인트에 대한 SLF 데이터는 N개의 컬러 정보를 담고 있는데, 포인트와 카메라 뷰포인트의 위치, 방향에 따라 중복된 컬러 정보를 가지고 있는 경우가 매우 많다. 그러므로 중복된 컬러 정보는 제외하고 대표적인 컬러 정보만을 선별하여 전송하는 것이 보다 효율적인 방법이 될 수 있다. 실시예들은 패치를 표현하는 카메라 뷰포인트들의 컬러 정보 중 서로 중복되지 않고 보다 차별적인 컬러 정보를 갖는 카메라 뷰포인트를 유의미한 카메라 뷰포인트로 판단하고, 선택하도록 한다.
전술한 제1방법과 같이, 사용된 카메라 뷰포인트의 카메라 matrix 와 패치의 normal vector 의 연산을 이용하여 패치의 방향과 가장 가깝게 마주보고 있는 카메라 뷰포인트를 선택한다. 이것을 패치의 컬러 정보를 가장 잘 표현하는 대표 카메라 뷰포인트 cam_rep로 정의한다. 그리고 나머지 N-1개의 카메라 뷰포인트들로부터 획득된 패치 컬러와 cam_rep으로부터 획득된 패치 컬러 간 차이를 각각 계산한다. cam_rep가 아닌 카메라 뷰포인트를 cam_k로 정의하고, 패치를 구성하는 모든 포인트들에 대하여 cam_rep와 cam_k 에서 가져온 컬러 값의 차이를 2400과 같이 계산한다.
패치 내 모든 포인트들의 2401값을 계산 후 이들의 평균 값 ave_C_diff_k 을 계산한다. cam_rep을 제외한 나머지 N - 1개의 카메라 뷰포인트들에 대해 ave_C_diff를 계산하고 이 값이 가장 큰 순서대로 M개의 카메라 뷰포인트들을 선택한다. ave_C_diff가 클수록 cam_rep에서 획득된 컬러와 차이가 많이 나는 컬러들을 해당 카메라 뷰포인트로부터 획득할 수 있고, 중복되지 않는 차별적인 컬러 정보로서 유의미한 정보가 될 수 있다고 판단할 수 있다. 여기에 추가적인 제약(constraint)으로 ave_C_diff가 일정 Threshold 이하인 카메라 뷰포인트는 선택하지 않도록 함으로써 비슷한 컬러 정보의 전송은 배제시킬 수 있다.
패치마다 결정되는 M의 값이 서로 다를 가능성이 높으며, M 값과 선택된 카메라 뷰포인트에 대한 정보는 패치마다 시그널링하여 전송한다.
실시예들에 따른 두 가지 방법 중 사용자의 선택에 따라 한 가지 방법을 사용하여 적용할 수 있으며, 사용된 선택방법은 시그널링하여 수신단에서 알 수 있도록 한다. 수신단에서의 사용자는 해당 SLF 시퀀스 컬러 데이터가 어떤 방식으로 선택되어 전송되었는지를 알 수 있고, 사용 목적에 따라 압축되어 온 SLF 컬러 정보를 선택적으로 활용할 수 있다.
또한, 패치 별 선택되는 카메라 뷰포인트 개수 M에 대해, M이 가질 수 있는 최대값 max_M은 2 <= max_M <= N 의 조건으로 해당 SLF 시퀀스에 대해 사전에 결정될 수 있다.
그 다음으로, 앞서 설명된 방법을 통해 선택된 패치 별 카메라 뷰포인트들의 컬러 정보를 이용하여 Texture 비디오를 생성하는 두 가지 방법을 제안한다.
Texture 비디오 생성 방법 1
입력된 SLF 시퀀스에 대해 생성되는 Texture 비디오의 개수는 선택된 카메라 뷰포인트의 개수들 중 최대값으로 결정되어야 한다. 그러므로 K= max(M)의 개수만큼 Texture 비디오를 생성하고, 패치 packing 과정에서 결정된 각 패치의 위치에는 선택된 카메라 뷰포인트로부터 획득된 컬러 정보를 mapping한다. K개의 Texture 비디오 각각에 패치 별로 컬러 정보를 mapping하는 순서는 패치 별로 카메라 뷰포인트가 선택되는 순서를 반영한다. 패치 별 카메라 뷰포인트 선택 방법 1이 사용된 경우에는, 패치로 향하는 방향이 가장 가까운 순서대로 카메라 뷰포인트들의 컬러를 각 Texture 비디오에 mapping 한다 (도24 참고). 예를 들어, 2402는 붉은색 컬러를 가진 패치 및 그 패치를 획득하는 카메라 뷰인트들이고, 2403은 파란색 컬러를 가진 패치 및 그 패치를 획득하는 카메라 뷰인트들이고, 2404는 노란색 컬러를 가진 패치 및 그 패치를 획득하는 카메라 뷰인트들이다.
패치 별 카메라 뷰포인트 선택 방법 2가 사용된 경우에는 cam_rep으로부터 획득된 컬러 정보를 첫번째 Texture 비디오에 mapping하고, 그 다음부터는 ave_C_diff가 큰 순서대로 해당 카메라 뷰포인트로부터 획득된 컬러 정보를 Texture 비디오에 mapping 한다. 이러한 순서를 적용하는 것은 Texture 비디오의 순서대로 컬러 정보의 중요도를 반영하기 위함이며, 이에 따라 첫번째 Texture 비디오가 가장 유의미한 컬러 정보를 담고 있다고 간주할 수 있다. 또한 이러한 순서의 적용은 수신단에서 사용자가 선택에 따라 모든 컬러를 복원할지 또는 중요도가 높은 일부의 컬러 데이터만 복원할지를 결정하는 데에 도움을 줄 수 있다.
패치 별로, 혹은 타일, 프레임 별로 선택된 카메라 뷰포인트 개수는 서로 다를 수 있기 때문에 경우에 따라 mapping될 컬러 정보가 없는 패치가 발생할 수 있다. 이 경우에는 해당 Texture 비디오에서의 패치 컬러는 0으로 채우도록 한다.
실시예들은 이와 같이 생성된 K개의 Texture 비디오를 2D 비디오 코덱을 이용하여 각각 압축하고, 전송하도록 한다.
도25는 실시예들에 따른 텍스쳐 비디오 생성 방법을 나타낸다.
Texture 비디오 생성 방법 2
Texture 비디오 생성 방법 1의 경우(2500), 생성된 Texture 비디오의 개수 K가 클수록 많은 코덱 인스턴스가 요구되므로, 본 Texture 비디오 생성 방법 2에서는 생성되는 Texture 비디오의 개수를 줄이고자 한다. 먼저 Texture 비디오 생성 방법 1에서와 동일하게 K개의 Texture 비디오를 생성한 후, 각 비디오의 각 프레임을 하나의 프레임으로 합쳐 1개의 Texture 비디오(2501)로 합성한다 (도25 참고). 가로/세로 방향으로 배치하는 비디오 프레임의 개수는 사용자의 지정에 따라 결정할 수 있다.
K개 비디오 index의 오름차순으로, 합성된 Texture 비디오 내에서는 레스터 스캔 순서(Raster scan order를 따라 각 프레임의 정보를 배치한다. 만약 합성된 Texture 비디오 내에 채울 수 없는 프레임 영역은 0으로 채우도록 한다. 이렇게 생성된 합성된 Texture 비디오는 한 개의 비디오 코덱을 이용하여 압축하고, 전송된다.
실시예들에 따른 시그널링 정보(파라미터 정보)를 통해 전송되는 Attribute 비디오가 합성된 비디오임을 알 수 있도록하며, 가로/세로 방향으로 존재하는 비디오의 개수를 시그널링함으로써 합성된 비디오의 구조에 대해 수신단에서 알 수 있도록 한다.
도26은 실시예들에 따른 V3C 비트스트림 구조를 나타낸다.
실시예들에 따른 포인트 클라우드 데이터 송신 방법/장치는 포인트 클라우드 데이터를 압축(인코딩)하고, 관련 파라미터 정보(도26 내지 39등)를 생성해서, 도26과 같은 비트스트림을 생성하여 전송할 수 있다.
실시예들에 따른 포인트 클라우드 데이터 수신 방법/장치는 도26과 같은 비트스트림을 수신하고, 비트스트림에 포함된 파라미터 정보에 기반하여 비트스트림에 포함된 포인트 클라우드 데이터를 디코딩할 수 있다.
실시예들에 따른 시그널링 정보(파라미터/메타데이터 등으로 지칭 가능함)는 실시예들에 따른 포인트 클라우드 데이터 송신 장치에서, 메타데이터 부화화부(메타데이터 부호화기 등으로 지칭 가능함)에 의해 인코딩되어 비트스트림에 포함되어 전송될 수 있다. 또한, 실시예들에 따른 포인트 클라우드 데이터 수신 장치에서, 메타데이터 복호화부(메타데이터 복호화기 등으로 지칭 가능함)에 의해 디코딩되어 포인트 클라우드 데이터의 디코딩 프로세스에 제공될 수 있다.
실시예들에 따른 송신기는 포인트 클라우드 데이터를 인코딩하여 비트스트림을 생성할 수 있다.
실시예들에 따른 비트스트림은 V3C 유닛(unit) 을 포함할 수 있다.
실시예들에 따른 수신기는 송신기가 전송한 비트스트림을 수신하고, 포인트 클라우드 데이터를 디코딩하고 복원할 수 있다. 이하에서, 실시예들에 따른 V3C unit의 구체적인 신택스 및 V3C unit에 포함된 엘리먼트들을 설명한다.
도27 내지 도39는 실시예들에 따른 포인트 클라우드 데이터 송신 방법/장치에 의해서 생성되는 V3C파라미터 세트 신택스이다.
도27-28은 실시예들에 따른 일반적인 V3C파라미터 세트이다.
파라미터 세트 아이디(vps_v3c_parameter_set_id): 다른 신택스 엘리먼트들에 의한 참조를 위한 VPS에 대한 식별자.
예약된 제로 비트(vps_reserved_zero_8bits): 비트스트림 내 0일 수 있다.
아틀라스 카운트(vps_atlas_count_minus1): 이 값에 1을 더하면, 현재 비트스트림 내 지원되는 아틀라스들의 총합. 0내지63의 값을 가질 수 있다.
아틀라스 아이디(vps_atlas_id[ k ]): 인덱스K를 가지는 아틀라스 아이디이다. 0내지63의 값을 가질 수 있다.
프레임 너비(vps_frame_width[ j ]): 아틀라스 아이디J를 가지는 아틀라스들을 위한 샘플들의 프레임 너비.
프레임 높이(vps_frame_height[ j ] ):아틀라스 아이디J를 가지는 아틀라스들을 위한 샘플들의 프레임 높이.
맵 카운트(vps_map_count_minus1[ j ]): 이 값에 1을 더하면, 아틀라스 아이디 J를 가지는 아틀라스를 위한 지오메트리 및 어트리뷰트 데이터를 인코딩하는데 상요되는 맵들의 개수를 나타낸다. 0내지15의 값을 가질 수 있다.
멀티플 맵 스트림 존재 플래그(vps_multiple_map_streams_present_flag[ j ]): 이 값이 0이면, 아틀라스 아이디J를 가지는 아틀라스에 대한 모든 지오메트리 또는 어트리뷰트 맵들이 싱글 지오메트리 또는 어트리뷰트 비디오 스트림 내 위치함을 나타낸다. 이 값이 1이면, 아틀라스 아이디J를 가지는 아틀라스에 대한 모든 지오메트리 또는 어트리뷰트 맵들이 개별적인 비디오 스트림들에 존재함을 나타낸다.
맵 앱솔루트 코딩 인에이블 플래그(vps_map_absolute_coding_enabled_flag[ j ][ i ]): 이 값이 1이면, 아틀라스 아이디 J를 가지는 아틀라스에 대한 인덱스I를 가지는 지오메트리 맵이 맵 프레딕션 없이 코딩됨을 나타낸다. 이 값이 0이면, 아틀라스 아이디 J를 가지는 아틀라스에 대한 인덱스I를 가지는 지오메트리 맵이 맵 코딩 이전에 먼저 코딩된 다른 것으로부터 먼저 예측됨을 나타낸다.
맵 프레딕터 인덱스 차이(vps_map_predictor_index_diff[ j ][ i ]): vps_map_absolute_coding_enabled_flag[ j ][ i ]이 0이면, 아틀라스 아이디 J를 가지는 아틀라스에 대한 인덱스 I를 가지는 지오메트리 맵의 예측기를 계산하는데 사용된다. 맵 프레딕터 인덱스는 다음과 같이 계산된다: MapPredictorIndex[ i ] = (i - 1) - vps_map_predictor_index_diff[ j ][ i ]
어실러리 비디오 존재 플래그(vps_auxiliary_video_present_flag[ j ]): 이 값이 1이면, 아틀라스 아이디 J를 가지는 아틀라스 내 패치에 대해 추가적인 정보, 즉 RAW 또는 EOM패치 타입들 관련된 정보가 개별적 비디오 스트림에 저장될 수 있음을 나타낸다. 개별적 비디오 스트림은 오실러리 비디오 스트림으로 지칭될 수 있다. 이 값이 0이면, 아틀라스 아이디 J를 가지는 아틀라스 내 패치에 대해 추가적인 정보, 즉 RAW 또는 EOM패치 타입들 관련된 정보가 오실러리 비디오 스트림에 저장되지 않음을 나타낸다.
어큐판시 비디오 존재 플래그(vps_occupancy_video_present_flag[ j ]): 이 값이 0이면, 아틀라스 아이디 J를 가지는 아틀라스가 관련된 어큐판시 비디오 데이터를 가지지 않음을 나타낸다. 이 값이 1이면, 아틀라스 아이디 J를 가지는 아틀라스가 관련된 어큐판시 비디오 데이터를 가짐을 나타낸다.
지오메트리 비디오 존재 플래그(vps_geometry_video_present_flag[ j ]): 이 값이 0이면, 아틀라스 아이디 J를 가지는 아틀라스가 관련 지오메트리 비디오 데이터를 가지지 않음을 나타낸다. 이 값이 1이면, 가짐을 나타낸다.
어틜뷰트 비디오 존재 플래그(vps_attribute_video_present_flag[ j ]):이 값이 0이면, 아틀라스 아이디 J를 가지는 아틀라스가 관련 어트리뷰트 비디오 데이터를 가지지 않음을 나타낸다. 이 값이 1이면, 가짐을 나타낸다.
도29는 실시예들에 따른 어트리뷰트 정보이다.
어트리뷰트 카운트(ai_attribute_count[ j ]):아틀라스 ID가 j인 아틀라스와 관련된 속성의 수를 나타낸다. ai_attribute_count[ j ]는 0에서 127까지의 범위일 수 있다.
어트리뷰트 타입 아이디(ai_attribute_type_id[ j ][ i ]): 아틀라스 ID가 j인 아틀라스에 대한 인덱스 i가 있는 속성 비디오 데이터 유닛의 속성 유형을 나타낸다.
V3C attribute types:
아이디가 0이면 식별자는 ATTR_TEXTURE이고, 타입은 Texture이다.
아이디가 1이면 식별자는 ATTR_MATERIAL_ID이고, 타입은 Material ID이다.
아이디가 2이면 식별자는 ATTR_TRANSPARENCY이고, 타입은 Transparency이다.
아이디가 3이면 식별자는 ATTR_REFLECTANCE이고, 타입은 Reflectance이다.
아이디가 4이면 식별자는 ATTR_NORMAL이고, 타입은 Normals이다.
아이디가 5-14이면 식별자는 ATTR_RESERVED이고, 타입은 Reserved이다.
아이디가 15이면 식별자는 ATTR_UNSPECIFIED이고, 타입은 Unspecified이다.
ATTR_TEXTURE : 볼륨메트릭 프레임의 텍스처 정보를 포함하는 속성을 나타낸다. 예를 들어 RGB(Red, Green, Blue) 색상 정보가 포함된 속성을 나타낼 수 있다.
ATTR_MATERIAL_ID : 볼륨메트릭 프레임에서 점의 재료 유형을 식별하는 추가 정보를 포함하는 속성을 나타낸다. 예를 들어, 재료 유형은 체적 프레임 내 점의 특성 또는 객체를 식별하기 위한 지표로 사용될 수 있다.
ATTR_TRANSPARENCY : 볼륨메트릭 프레임의 각 점과 연관된 투명도 정보가 포함된 속성을 나타낸다.
ATTR_REFLECTANCE : 볼륨메트릭 프레임의 각 점과 연관된 반사율 정보가 포함된 속성을 나타낸다.
ATTR_NORMAL : 볼륨메트릭 프레임의 각 점과 관련된 단위 벡터 정보를 포함하는 속성을 나타낸다. 단위 벡터는 한 점에서 표면에 대한 수직 방향(즉, 점이 향하는 방향)을 지정한다. 이 속성 유형을 갖는 속성 프레임은 2와 동일한 ai_attribute_dimension_minus1을 가질 수 있다. 이 속성 유형을 갖는 속성 프레임의 각 채널은 단위 벡터(x, y, z)의 하나의 구성요소를 포함할 수 있다. 여기서 첫 번째 구성요소는 x 좌표를 포함하고, 두 번째 구성요소는 y 좌표를 포함하고 세 번째 구성요소는 z 좌표를 포함한다.
ATTR_UNSPECIFIED 특정 의미 없는 값을 나타낼 수 있다. ATTR_RESERVED로 표시된 값은 ISO/IEC의 향후 사용을 위해 예약되어 있으며 이 버전의 이 문서를 준수하는 비트스트림에 존재하지 않을 수 있다.
어트리뷰트 코덱 아이디(ai_attribute_codec_id[ j ][ i ]): 아틀라스 ID가 j인 아틀라스에 대한 인덱스 i가 있는 속성 비디오 데이터를 압축하는 데 사용되는 코덱의 식별자를 나타낸다. ai_attribute_codec_id[ j ][ i ]는 0에서 255까지의 범위에 있을 수 있다.
오실러리 어트리뷰트 코덱 아이디(ai_auxiliary_attribute_codec_id[ j ][ i ]): RAW 및/또는 EOM 코딩된 포인트가 아틀라스 ID j를 갖는 아틀라스에 대한 보조 비디오 스트림으로 인코딩될 때 속성 i의 RAW 및/또는 EOM 코딩된 포인트에 대한 속성 비디오 데이터를 압축하는 데 사용되는 코덱의 식별자를 나타낸다. ai_auxiliary_attribute_codec_id[ j ][ i ]는 0에서 255까지의 범위에 있을 수 있다. 이 값이 존재하지 않을 때 ai_auxiliary_attribute_codec_id[ j ][ i ]의 값은 ai_attribute_codec_id[ j ][ i ]와 동일한 것으로 추론된다.
어트리뷰트 맵 앱솔루트 코딩 존재 플래그(ai_attribute_map_absolute_coding_persistence_flag[ j ][ i ]): 이 값이 1이면, 인덱스 i가 있는 속성에 대해 아틀라스 ID j가 있는 아틀라스에 해당하는 모든 속성 맵이 어떤 형태의 맵 예측 없이 코딩됨을 나타낸다. ai_attribute_map_absolute_coding_persistence_flag[ j ][ i ]가 0이면 아틀라스 ID가 j인 아틀라스에 해당하는 인덱스 i가 있는 속성의 속성 맵이 아틀라스가 있는 아틀라스의 지오메트리 구성요소에 사용된 것과 동일한 맵 예측 방법을 사용해야 함을 나타낸다. ai_attribute_map_absolute_coding_persistence_flag[ j ][ i ]가 존재하지 않는 경우, 그 값은 1과 동일한 것으로 유추될 수 있다.
속성의 특정 맵이 예측을 포함하거나 포함하지 않고 코딩되어야 하는지 여부를 나타내는 3D 배열 AttributeMapAbsoluteCodingEnabledFlag는 다음과 같이 획득된다:
if( ai_attribute_map_absolute_coding_persistance_flag[ j ][ i ] == 1) {
for( k = 0; k < vps_map_count_minus1[ j ]; k++ )
AttributeMapAbsoluteCodingEnabledFlag[ j ][ i ][ k ] = 1
}
else{
for( k = 0; k < vps_map_count_minus1[ j ]; k++ )
AttributeMapAbsoluteCodingEnabledFlag[ j ][ i ][ k ] =
vps_map_absolute_coding_enabled_flag[ j ][ i ]
}
어트리뷰트 디멘션(ai_attribute_dimension_minus1[ j ][ i ]): 이 값에 1을 더하면, 아틀라스 ID가 j인 아틀라스에 대해 인덱스 i가 있는 속성의 총 차원 수(즉, 채널 수)를 나타낸다. ai_attribute_dimension_minus1[ j ][ i ]는 0에서 63까지의 범위에 있을 수 있다.
어트리뷰트 디멘션 파티션들(ai_attribute_dimension_partitions_minus1[ j ][ i ]): 이 값에 1을 더하면, 인덱스 i가 있는 속성의 속성 채널이 아틀라스 ID j가 있는 아틀라스에 대해 그룹화되어야 하는 파티션 그룹의 수를 나타낸다. ai_attribute_dimension_partitions_minus1[ j ][ i ]는 0에서 63 사이의 범위에 있을 수 있다.
어트리뷰트 파티션 채널들(ai_attribute_partition_channels_minus1[ k ][ i ][ j ]): 이 값에 1을 더하면, 아틀라스 ID가 k인 아틀라스에 대해 인덱스 i가 있는 속성의 인덱스 j가 있는 차원 파티션 그룹에 할당된 채널 수를 나타낸다. ai_attribute_partition_channels_minus1[ k ][ i ][ j ]는 모든 차원 파티션 그룹에 대해 0에서 ai_attribute_dimension_minus1[ k ][ i ]까지의 범위에 있을 수 있다.
어트리뷰트 비트 뎁스(ai_attribute_2d_bit_depth_minus1[ j ][ i ]): 이 값에 1을 더하면, 아틀라스 ID가 j인 아틀라스에 대해 속성 인덱스가 i인 모든 속성 비디오가 변환되어야 하는 2D 비트 깊이를 나타냅니다. ai_attribute_2d_bit_depth_minus1[ j ][ i ]는 0에서 31까지의 범위에 있을 수 있다.
어트리뷰트 MSB 얼라인 플래그(ai_attribute_MSB_align_flag[ j ][ i ]): 아틀라스 ID j를 가진 아틀라스에 대해 속성 인덱스 i를 가진 디코딩된 속성 비디오 샘플이 속성 비트 깊이의 샘플로 변환되는 방법을 나타낸다.
어트리뷰트 카운트(ai_attribute_count[ j ]): 아틀라스 ID가 j인 아틀라스와 연결된 속성의 수를 나타낸다. ai_attribute_count[ j ]는 0에서 127까지의 범위에 있을 수 있다.
어트리뷰트 SLF 카메라뷰 선택 타입(ai_attribute_slf_cameraview_selection_type[ j ]_: SLF 데이터의 컬러 정보 압축 여부와, SLF 데이터 압축 시 사용된 패치 별 카메라 뷰포인트 선택 방법을 나타낸다. 이 값이 0인 경우, SLF 데이터가 아닌 일반적인 포인트 클라우드 데이터의 attribute 정보가 전송됨을 나타낸다. 이 값이 1 내지 3인 경우는 SLF 데이터로부터 얻은 Attribute 정보가 전송됨을 나타내는데, 1이면 Attribute 비디오 생성 시 패치 별 카메라 뷰포인트 선택 방법으로 실시예들에 따른 패치 별 카메라 뷰포인트 선택 방법 1이 사용되었음을 나타내고, 2이면 실시예들에 따른 패치 별 카메라 뷰포인트 선택 방법 2가 사용되었음을 나타내고, 3이면 패치 별 카메라 뷰포인트를 선택하지 않고 모든 카메라 뷰포인트로부터 얻은 Attribute 정보를 전송함을 나타낸다.
어트리뷰트 머지된 비디오 플래그(ai_attribute_merged_video_flag[ j ][ i ]): 이 값이 1이면 전송되는 Attribute 비디오가 실시예들에 따른 Texture 비디오 생성 방법 2를 사용하여 생성된 합성된 Attribute 비디오임을 나타낸다. 이 값이 0이면 Texture 비디오 생성 방법 1에서 설명된 대로 생성된 K개의 Attribute 비디오가 각각 전송됨을 나타낸다.
컬럼 내 어트리뷰트 머지된 어트리뷰트들 개수(ai_attribute_number_of_merged_attributes_in_column[ j ][ i ]): ai_attribute_merged_video_flag[ j ][ i ] = 1인 경우, 합성된 Attribute 비디오의 합성 구조를 나타내기 위해 사용되는 신택스이다. 도30과와 같은 방법으로 Attribute 비디오가 합성되어 하나의 Merged attribute 비디오로 전송되는 경우, ai_attribute_number_of_merged_attributes_in_column[ j ][ i ] = 2로 시그널링된다.
로우 내 어트리뷰트 머지된 어트리뷰들 개수(ai_attribute_number_of_merged_attributes_in_row[ j ][ i ] ai_attribute_merged_video_flag[ j ][ i ] = 1인 경우, 합성된 Attribute 비디오의 합성 구조를 나타내기 위해 사용되는 syntax이다. Figure 6와 같은 방법으로 Attribute 비디오가 합성되어 하나의 Merged attribute 비디오로 전송되는 경우, ai_attribute_number_of_merged_attributes_in_row[ j ][ i ] = 3으로 시그널링된다.
도30은 실시예들에 따른 비디오 합성 구조를 나타낸다.
도30은 6개의 어트리뷰트 비디오가 하나의 머지된 어트리뷰트 비디오로 합성되어 전송되는 경우, 비디오 합성 구조 예시이다.
도31-32는 실시예들에 따른 아틀라스 시퀀스 파라미터 세트를 나타낸다.
아틀라스 시퀀스 파라미터 세트 아이디(asps_atlas_sequence_parameter_set_id): 다른 구문 요소에서 참조할 수 있도록 아틀라스 시퀀스 매개변수 세트에 대한 식별자를 제공한다.
프레임 너비(asps_frame_width): 샘플의 정수 개수로 아틀라스 프레임 너비를 나타낸다. 여기서 샘플은 비디오 구성요소의 루마 샘플에 해당한다. asps_frame_width의 값이 vps_frame_width[ j ]의 값과 같아야 하는 V3C 비트스트림 적합성의 요구사항을 따를 수 있다. 여기서 j는 현재 지도의 ID이다.
프레임 높이(asps_frame_height): 샘플의 정수 개수로 아틀라스 프레임 높이를 나타낸다. 여기서 샘플은 비디오 구성요소의 루마 샘플에 해당한다. asps_frame_height의 값이 vps_frame_height[ j ]의 값과 같아야 하는 V3C 비트스트림 적합성의 요구사항을 따를 수 있다. 여기서 j는 현재 지도의 ID이다.
지오메트리 3d 비트 뎁스(asps_geometry_3d_bit_depth_minus1): 이 값에 1을 더하면, 재구성된 체적 콘텐츠의 기하 좌표의 비트 깊이를 나타낸다. asps_geometry_3d_bit_depth_minus1은 0에서 31까지의 범위에 있을 수 있다.
지오메트리 2d 비트 뎁스(asps_geometry_2d_bit_depth_minus1): 이 값에 1을 더하면 2D 이미지에 투영될 때 지오메트리의 비트 깊이를 나타낸다. asps_geometry_2d_bit_depth_minus1은 0에서 31까지의 범위에 있을 수 있다.
어트리뷰트 사용 카메라뷰 개수 플래그(asps_attribute_use_fixed_number_of_cameraview_flag): 실시예들에 따른 방법을 이용하여 패치 별 카메라 뷰포인트를 선택하고 이를 통해 Attribute 비디오를 생성하는 경우, 해당 시퀀스 내 모든 패치에 대해 선택되는 카메라 뷰포인트의 개수가 동일하면 이 값은 1을 가질 수 있다. 반면, 프레임 혹은 타일 혹은 패치 단위로, 선택되는 패치 별 카메라 뷰포인트의 개수가 서로 다르게 적용되면 이 값은 0을 가질 수 있다.
asps_attribute_selected_cameraview_count 해당 시퀀스 내 모든 패치에 대해 동일한 개수의 카메라 뷰포인트를 선택하도록 하는 경우, 그 개수의 값을 시그널링한다. 만약 이 값이 7이면, A. 1)에서 설명된 M = 7이며, 시퀀스 내 모든 패치는 각각 7개의 카메라 뷰포인트를 선택한다.
맥스 아틀라스 프레임 오더 카운트(asps_log2_max_atlas_frame_order_cnt_lsb_minus4): 다음과 같이 아틀라스 프레임 순서 카운트에 대한 디코딩 프로세스에서 사용되는 변수 MaxAtlasFrmOrderCntLsb의 값을 나타낸다: MaxAtlasFrmOrderCntLsb = 2asps_log2_max_atlas_frame_order_cnt_lsb_minus4 + 4. asps_log2_max_atlas_frame_order_cnt_lsb_minus4의 값은 0에서 12까지의 범위에 있을 수 있다.
맥스 디코딩 아틀라스 프레임 버퍼링(asps_max_dec_atlas_frame_buffering_minus1):
이 값에 1을 더하면, CAS에 대해 디코딩된 아틀라스 프레임 버퍼의 최대 필수 크기를 아틀라스 프레임 저장 버퍼 단위로 지정한다. asps_max_dec_atlas_frame_buffering_minus1의 값은 0에서 15까지의 범위에 있을 수 있다.
롱 텀 레퍼런스 아틀라스 프레임 플래그(asps_long_term_ref_atlas_frames_flag): 이 값이 0이면, CAS에서 코딩된 아틀라스 프레임의 인터 예측에 장기 참조 아틀라스 프레임이 사용되지 않음을 나타낸다. 1과 동일한 asps_long_term_ref_atlas_frames_flag는 CAS에서 하나 이상의 코딩된 아틀라스 프레임의 인터 예측을 위해 장기 참조 아틀라스 프레임이 사용될 수 있음을 나타낸다.
레퍼런스 아틀라스 프레임 리스트 개수(asps_num_ref_atlas_frame_lists_in_asps): 아틀라스 시퀀스 매개변수 세트에 포함된 ref_list_struct( rlsIdx ) 구문 구조의 수를 나타낸다. asps_num_ref_atlas_frame_lists_in_asps의 값은 0에서 64까지의 범위에 있을 수 있다.
오리엔테이션 플래그(asps_use_eight_orientations_flag): 0이면, 타일 ID가 i인 타일에서 인덱스 j가 있는 패치에 대한 패치 방향 인덱스, pdu_orientation_index[ i ][ j ]가 0에서 1(포함)의 범위에 있음을 나타낸다. 1과 동일한 asps_use_eight_orientations_flag는 pdu_orientation_index[ i ][ j ]가 0에서 7까지의 범위에 있음을 나타낸다.
프로젝션 인에이블 플래그(asps_extended_projection_enabled_flag): 0이면, 현재 아틀라스 타일에 대해 패치 투영 정보가 신호되지 않음을 나타낸다. 1이면, asps_extended_projection_enabled_flag는 현재 아틀라스 타일에 대해 패치 투영 정보가 시그널링되는 것을 나타낸다. asps_extended_projection_enabled_flag가 없으면 해당 값은 0으로 유추된다.
프로젝션 맥스 개수(asps_max_number_projections_minus1): 이 값에 1을 더하면, 타일 ID가 i인 타일에서 인덱스 j가 있는 패치에 대해 패치 투영 ID 구문 요소 pdu_projection_id[ i ][ j ]에 대해 표시될 수 있는 최대값을 나타낸다. asps_max_number_projections_minus1이 없으면 그 값은 5와 같은 것으로 유추된다.
노멀 축 리미트 양자화 인에이블 플래그(asps_normal_axis_limits_quantization_enabled_flag): 1이면, 패치 데이터 단위, 병합 패치 데이터 단위 또는 패치 간 데이터 단위의 법선 축 관련 요소를 양자화하기 위해 양자화 매개변수가 신호를 받고 사용되어야 함을 나타낸다. asps_normal_axis_limits_quantization_enabled_flag가 0과 같으면 패치 데이터 단위, 병합 패치 데이터 단위 또는 패치 간 데이터 단위의 법선 축 관련 요소에 양자화가 적용되지 않는다.
노멀 축 맥스 델타 값 인에이블 플래그(asps_normal_axis_max_delta_value_enabled_flag): 1이면, 인덱스 j가 있는 프레임에서 인덱스 i가 있는 패치의 기하 정보에 존재할 수 있는 노멀 축의 최대 노미널 시프트 값이 각 패치 데이터 단위, 병합 패치 데이터에 대한 비트스트림에 표시될 것임을 나타낸다. 단위 또는 패치간 데이터 단위. asps_normal_axis_max_delta_value_enabled_flag가 0이면 인덱스 j가 있는 프레임에서 인덱스 i가 있는 패치의 기하 정보에 존재할 수 있는 노멀 축의 최대 노미널 이동 값은 각 패치 데이터 단위, 병합 패치 데이터 유닛, 또는 인터 패치 데이터 유닛에 대한 비트스트림에 표시되지 않을 수 있다.
패치 프리시던스 오더 플래그(asps_patch_precedence_order_flag): 아틀라스 샘플을 패치에 할당하는 데 사용되는 패치 우선 순위를 나타낸다. asps_patch_precedence_order_flag가 1이면 현재 아틀라스에 대한 패치 우선 순위가 패치의 디코딩 순서와 동일하다는 것을 나타낸다. 0과 동일한 asps_patch_precedence_order_flag는 현재 아틀라스에 대한 패치 우선 순위가 패치의 디코딩 순서의 반대임을 나타낸다.
패치 패킹 블록 사이즈(asps_log2_patch_packing_block_size): 다음과 같이 아틀라스 내 패치의 수평 및 수직 배치에 사용되는 PatchPackingBlockSize 변수 값을 나타낸다: PatchPackingBlockSize = 2 asps_log2_patch_packing_block_size.
asps_log2_patch_packing_block_size의 값은 0에서 7까지의 범위에 있을 수 있다.
패치 사이즈 양자화 존재 플래그(asps_patch_size_quantizer_present_flag): 1이면, 패치 크기 양자화 매개변수가 아틀라스 타일 헤더에 있음을 나타낸다. asps_patch_size_quantizer_present_flag가 0과 같으면 패치 크기 양자화 매개변수가 존재하지 않을 수 있다.
맵 카운트(asps_map_count_minus1): 이 값에 1을 더하면 현재 아틀라스의 지오메트리 및 속성 데이터를 인코딩하는 데 사용할 수 있는 맵의 수를 나타낸다. asps_map_count_minus1은 0에서 15까지의 범위에 있을 수 있다. 이 문서에 대한 비트스트림 준수의 요구 사항은 asps_map_count_minus1이 vps_map_count_minus1[ atlasID ]와 동일하며, 여기서 atlasID는 현재 지도의 지도 ID이다.
픽셀 디인터리빙 인에이블 플래그(asps_pixel_deinterleaving_enabled_flag): 1과 같으면 디코딩된 지오메트리 및 속성 데이터가 재구성 동안 추가 공간 보간 프로세스가 필요할 수 있음을 나타낸다. 0과 동일한 asps_pixel_deinterleaving_enabled_flag는 디코딩된 지오메트리 및 속성 데이터에 추가 공간 보간 프로세스가 필요하지 않음을 나타낸다.
맵 픽셀 디인터리빙 플래그(asps_map_pixel_deinterleaving_flag[ i ]): 1과 같음은 현재 지도에서 인덱스 i가 있는 지도에서 투영된 패치의 관련 지오메트리 및 속성 데이터에 대해 추가 공간 보간 프로세스를 수행해야 함을 나타낸다. 0과 동일한 asps_map_pixel_deinterleaving_flag[ i ]는 추가 공간 보간 프로세스가 수행되지 않음을 나타낸다. 존재하지 않는 경우 asps_map_pixel_deinterleaving_flag[ i ]의 값은 0으로 유추된다.
로우 패치 인에이블 플래그(asps_raw_patch_enabled_flag): 1이면, 현재 아틀라스에 대한 디코딩된 지오메트리 및 속성 비디오에 RAW 코딩된 포인트와 관련된 정보가 포함되어 있음을 나타낸다. 0과 동일한 asps_raw_patch_enabled_flag는 디코딩된 지오메트리 및 속성 비디오가 RAW 코딩된 포인트와 관련된 정보를 포함하지 않음을 나타낸다.
EOM 패치 인에이블 플래그(asps_eom_patch_enabled_flag): 1이면, 현재 아틀라스에 대한 디코딩된 점유 비디오가 두 개의 깊이 맵 사이의 중간 깊이 위치가 점유되는지 여부와 관련된 정보를 포함한다는 것을 나타낸다. 0과 동일한 asps_eom_patch_enabled_flag는 디코딩된 점유 비디오가 2개의 깊이 맵 사이의 중간 깊이 위치가 점유되는지 여부와 관련된 정보를 포함하지 않음을 나타낸다. asps_eom_patch_enabled_flag가 1과 같으면 oi_lossy_occupancy_compression_threshold[ atlasID ]가 0과 같아야 하는 비트스트림 적합성의 요구 사항을 따를 수 있다.
EOM 픽스 비트 카운트(asps_eom_fix_bit_count_minus1): 이 값에 1을 더하면, EOM 코드워드의 비트 단위 크기를 나타낸다. asps_eom_fix_bit_count_minus1은 0에서 Min(15, oi_occupancy_2d_bit_depth_minus1[ atlasID ] - 1) 범위에 있을 수 있다. 여기서 atlasID는 현재 지도의 지도 ID이다.
오실러리 비디오 인에이블 플래그(]asps_auxiliary_video_enabled_flag): 1이면, RAW 및 EOM 패치 유형과 관련된 정보가 보조 비디오 하위 비트스트림에 배치될 수 있음을 나타낸다. 0과 동일한 asps_auxiliary_video_enabled_flag는 RAW 및 EOM 패치 유형과 관련된 정보가 기본 비디오 하위 비트스트림에만 배치될 수 있음을 나타낸다.
vps_auxiliary_video_present_flag[ atlasID ](여기서 atlasID는 현재 아틀라스의 아틀라스 ID임)가 0이면 asps_auxiliary_video_enabled_flag가 0과 같은 비트스트림 적합성의 요구사항을 따를 수 있다.
포인트 로컬 복원 인에이블 플래그(asps_plr_enabled_flag): 1이면, 포인트 로컬 재구성 모드 정보가 현재 아틀라스에 대한 비트스트림에 존재할 수 있음을 나타낸다. 0과 동일한 asps_plr_enabled_flag는 포인트 로컬 재구성 모드와 관련된 정보가 현재 아틀라스에 대한 비트스트림에 존재하지 않음을 나타낸다.
asps_pixel_deinterleaving_enabled_flag가 1과 같을 때 asps_plr_enabled_flag가 0과 같아야 하는 것이 비트스트림 적합성의 요구 사항일 수 있다.
파라미터 존재 플래그(asps_vui_parameters_present_flag): 1이면, vui_parameters() 구문 구조가 존재함을 나타낸다. 0과 동일한 asps_vui_parameters_present_flag는 vui_parameters() 구문 구조가 존재하지 않음을 나타낸다.
확장 존재 플래그(asps_extension_present_flag): 1이면, 구문 요소 asps_vpcc_extension_present_flag 및 asps_extension_7bits가 atlas_sequence_parameter_set_rbsp 구문 구조에 존재함을 나타낸다. 0과 동일한 asps_extension_present_flag는 구문 요소 asps_vpcc_extension_present_flag 및 asps_extension_7bits가 존재하지 않음을 나타낸다.
VPCC확장 존재 플래그(asps_vpcc_extension_present_flag): 1이면, asps_vpcc_extension() 구문 구조가 atlas_sequence_parameter_set_rbsp 구문 구조에 있음을 나타낸다. 0과 동일한 asps_vpcc_extension_present_flag는 이 구문 구조가 존재하지 않음을 나타낸다. 존재하지 않을 때 asps_vpcc_extension_present_flag의 값은 0과 같은 것으로 추론된다.
확장 비트(asps_extension_7bits): 0이면, ASPS RBSP 구문 구조에 asps_extension_data_flag 구문 요소가 존재하지 않음을 나타낸다. 존재할 때, asps_extension_7bits는 이 문서의 이 버전을 따르는 비트스트림에서 0일 수 있다. 0이 아닌 asps_extension_7bits의 값은 ISO/IEC에서 향후 사용을 위해 예약될 수 있다. 디코더는 asps_extension_7bits의 값을 0이 아닌 다른 값으로 허용해야 하며 ASPS NAL 단위의 모든 asps_extension_data_flag 구문 요소를 무시할 수 있다. 존재하지 않는 경우 asps_extension_7bits의 값은 0과 같은 것으로 유추된다.
확장 데이터 플래그(asps_extension_data_flag): 어떤 값을 가질 수 있다. 그것의 존재와 가치는 이 문서의 이 버전에 지정된 프로파일에 대한 디코더 적합성에 영향을 미치지 않는다. 이 문서의 이 버전을 따르는 디코더는 모든 asps_extension_data_flag 구문 요소를 무시할 수 있다.
도33은 실시예들에 다른 아틀라스 프레임 파라미터 세트를 나타낸다.
아틀라스 프레임 파라미터 세트 아이디(afps_atlas_frame_parameter_set_id): 다른 구문 요소에서 참조하기 위해 아틀라스 프레임 매개변수 세트를 식별한다.
아틀라스 시퀀스 파라미터 세트 아이디(afps_atlas_sequence_parameter_set_id): 액티브 아틀라스 시퀀스 매개변수 세트에 대한 asps_atlas_sequence_parameter_set_id 값을 나타낸다.
어트리뷰트 사용 고정 카메라뷰 개수(afps_attribute_use_fixed_number_of_cameraview_flag) 실시예들에 따른 방법을 이용하여 패치 별 카메라 뷰포인트를 선택하고 이를 통해 Attribute 비디오를 생성하는 경우, 해당 프레임 내 모든 패치에 대해 선택되는 카메라 뷰포인트의 개수가 동일하면 이 값은 1이 된다. 반면, 타일 혹은 패치 단위로, 선택되는 패치 별 카메라 뷰포인트의 개수가 서로 다르게 적용되면 이 값은 0이 된다.
어트리뷰트 선택된 카메라뷰 카운트(afps_attribute_selected_cameraview_count): 해당 프레임 내 모든 패치에 대해 동일한 개수의 카메라 뷰포인트를 선택하도록 하는 경우, 그 개수의 값을 시그널링한다. 만약 이 값이 7이면, 해당 프레임에 대해서는 상술한 바와 같이 M = 7이며 프레임 내 모든 패치는 각각 7개의 카메라 뷰포인트를 선택한다.
아웃풋 플래그 존재 플래그(afps_output_flag_present_flag): 1이면, ath_atlas_output_flag 신택스 요소가 연관된 타일 헤더에 존재함을 나타낸다. 0과 동일한 afps_output_flag_present_flag는 ath_atlas_output_flag 신택스 요소가 연관된 타일 헤더에 존재하지 않음을 나타낸다.
레퍼런스 인덱스 개수(afps_num_ref_idx_default_active_minus1): 이 값에 1을 더하면, ath_num_ref_idx_active_override_flag가 0인 타일에 대한 변수 NumRefIdxActive의 추론된 값을 나타낸다. afps_num_ref_idx_default_active_minus1의 값은 0에서 14까지의 범위를 가질 수 있다.
추가적인 길이(afps_additional_lt_afoc_lsb_len): 참조 아틀라스 프레임 목록에 대한 디코딩 프로세스에 사용되는 MaxLtAtlasFrmOrderCntLsb 변수의 값을 다음과 같이 지정할 수 있다: MaxLtAtlasFrmOrderCntLsb = 2 * ( asps_log2_max_atlas_frame_order_cnt_lsb_minus4 +4 + afps_additional_lt_afoc_lsb_len). afps_additional_lt_afoc_lsb_len의 값은 0에서 28(asps_log2_max_atlas_frame_order_cnt_lsb_minus4 포함)의 범위에 있을 수 있다.
asps_long_term_ref_atlas_frames_flag가 0과 같을 때, afps_additional_lt_afoc_lsb_len의 값은 0과 같을 수 있다.
LOD 모드 인에이블 플래그(afps_lod_mode_enabled_flag): 1이면 LOD 매개변수가 패치에 존재할 수 있음을 나타낸다. 0과 동일한 afps_lod_mode_enabled_flag는 LOD 매개변수가 패치에 존재하지 않아야 함을 나타낸다.
로우3D 오프셋 비트 카운트 모드 플래그(afps_raw_3d_offset_bit_count_explicit_mode_flag): 1이면, tileID와 동일한 타일 ID를 갖는 타일 내 인덱스 p를 갖는 패치 에 대한 rpdu_3d_offset_u[ tileID ][ p ], rpdu_3d_offset_v[ tileID ][ p ] 및 rpdu_3d_offset_d[ tileID ][ p ]의 고정 길이 표현 내 비트들이 afps_atlas_frame_parameter_set_id를 참조하는 아틀라스 타일 헤더의 ath_raw_3d_offset_axis_bit_count_minus1에 의해 명시적으로 코딩됨을 나타낸다. 0과 동일한 afps_raw_3d_offset_bit_count_explicit_mode_flag는 ath_raw_3d_offset_axis_bit_count_minus1의 값이 암시적으로 파생됨을 나타낸다.
확장 존재 플래그(afps_extension_present_flag): 1이면, afps_extension_8bits 구문 요소가 atlas_frame_parameter_set_rbsp 구문 구조에 존재함을 나타낸다. 0과 동일한 afps_extension_present_flag는 구문 요소 afps_extension_8bits가 존재하지 않음을 나타낸다. afps_extension_present_flag의 값은 이 문서의 이 버전에서 0일 수 있다.
확장 비트(afps_extension_8bits): 0이면, AFPS RBSP 신택스 구조에 afps_extension_data_flag 신택스 요소가 존재하지 않음을 나타낸다. 존재하는 경우 afps_extension_8bits는 이 버전의 이 문서를 준수하는 비트스트림에서 0일 수 있다. 0이 아닌 afps_extension_8bits의 값은 ISO/IEC에서 향후 사용을 위해 예약될 수 있다. 디코더는 afps_extension_8bits 값을 0 이외의 값으로 허용하고 AFPS NAL 단위의 모든 afps_extension_data_flag 구문 요소를 무시할 수 있다. 존재하지 않는 경우 afps_extension_8bits의 값은 0과 같은 것으로 유추된다.
확장 데이터 플래그(afps_extension_data_flag): 어떤 값을 가질 수 있다. 그것의 존재와 가치는 이 문서의 이 버전에 지정된 프로파일에 대한 디코더 적합성에 영향을 미치지 않는다. 이 문서의 이 버전을 따르는 디코더는 모든 afps_extension_data_flag 구문 요소를 무시할 수 있다.
도34-35는 실시예들에 따른 아틀라스 프레임 타일 정보를 나타낸다.
아틀라스 프레임 내 싱글 타일 플래그(afti_single_tile_in_atlas_frame_flag): 1이면, AFPS를 참조하는 각 아틀라스 프레임에 타일이 하나만 있음을 나타낸다. 0과 동일한 afti_single_tile_in_atlas_frame_flag는 AFPS를 참조하는 각 아틀라스 프레임에 하나 이상의 타일이 있을 수 있음을 나타낸다.
유니폼 파티션 스페이싱 플래그(afti_uniform_partition_spacing_flag): 1이면, 아틀라스의 타일 파티셔닝이 아틀라스 프레임 전체에 열 및 행 파티션 경계를 균일하게 분배하는 방법을 사용하는 것을 나타낸다. 이러한 경계에 해당하는 정보는 각각 afti_partition_cols_width_minus1 및 afti_partition_rows_height_minus1 구문 요소를 사용하여 시그널링됩니다. 0과 동일한 afti_uniform_partition_spacing_flag는 아틀라스의 타일 파티셔닝이 아틀라스 프레임 전체에 균일하게 분배되거나 분배되지 않을 수 있는 열 및 행 파티션 경계를 생성할 수 있는 방법을 사용하는 것을 나타낸다. 이 경우, 이러한 경계는 구문 요소 afti_num_partition_columns_minus1 및 afti_num_partition_rows_minus1 및 구문 요소 쌍 afti_partition_column_width_minus1[ i ] 및 afti_partition_row_height_minus1[ i ] 목록을 사용하여 시널링된다. 존재하지 않을 때, afti_uniform_partition_spacing_flag의 값은 1과 같은 것으로 추론된다.
파티션 컬럼 너비(afti_partition_cols_width_minus1): 이 값에 1을 더하면, afti_uniform_partition_spacing_flag가 1일 때 아틀라스 프레임의 가장 오른쪽 타일 파티션 열을 제외한 타일 파티션 열의 너비를 나타낸다. 존재하지 않는 경우, afti_partition_cols_width_minus1의 값은 asps_frame_width / 64-1과 동일한 것으로 유추된다.
파티션 로우 높이(afti_partition_rows_height_minus1): 이 값에 1을 더하면, afti_uniform_partition_spacing_flag가 1일 때 아틀라스 프레임의 하단 타일 파티션 행을 제외한 타일 파티션 행의 높이를 64 샘플 단위로 나타낸다. 포함한. 존재하지 않는 경우, afti_partition_rows_height_minus1의 값은 asps_frame_height/64-1과 동일한 것으로 유추된다.
파티션 컬럼 개수(afti_num_partition_columns_minus1): 이 값에 1을 더하면, afti_uniform_partition_spacing_flag가 0과 같을 때 아틀라스 프레임을 분할하는 데 사용되는 타일 파티션 열의 수를 나타낸다. afti_num_partition_columns_minus1의 값은 0에서 asps_frame_width / 64-1까지의 범위에 있을 수 있다. 지정된 afti_single_tile_in_atlas_frame_flag가 1과 같으면 afti_num_partition_columns_minus1의 값은 0과 같은 것으로 추론된다.
파티션 로우 개수(afti_num_partition_rows_minus1): 이 값에 1을 더하면, afti_uniform_partition_spacing_flag가 0과 같을 때 아틀라스 프레임을 분할하는 데 사용되는 타일 파티션 행의 수를 나타낸다. afti_num_partition_rows_minus1의 값은 0에서 asps_frame_height / 64-1(포함)의 범위에 있을 수 있다. afti_single_tile_in_atlas_frame_flag가 1과 같으면 afti_num_partition_rows_minus1의 값이 0과 같은 것으로 추론된다.
NumPartitionsInAtlasFrame 변수는 NumPartitionColumns * NumPartitionRows와 동일하게 설정된다.
afti_single_tile_in_atlas_frame_flag가 0과 같을 때 NumPartitionsInAtlasFrame은 1보다 클 수 있다.
파티션 컬럼 너비(afti_partition_column_width_minus1[ i ]): 이 값에 1을 더하면, i번째 타일 파티션 열의 너비를 64개 샘플 단위로 나타낸다.
파티션 로우 높이(afti_partition_row_height_minus1[ i ]): 이 값에 1을 더하면, i번째 타일 파티션 행의 높이를 64개 샘플 단위로 나타낸다.
타일 당 싱글 파티션 플래그(afti_single_partition_per_tile_flag): 1이면, 이 AFPS를 참조하는 각 타일이 하나의 타일 파티션을 포함하는 것을 나타낸다. 0과 동일한 afti_single_partition_per_tile_flag는 이 AFPS를 참조하는 타일이 하나 이상의 타일 파티션을 포함할 수 있음을 나타낸다. 존재하지 않을 때, afti_single_partition_per_tile_flag의 값은 1과 같은 것으로 추론된다.
아틀라스 프레임 내 타일들 개수(afti_num_tiles_in_atlas_frame_minus1): 이 값에 1을 더하면, AFPS를 참조하는 각 아틀라스 프레임의 타일 수를 나타낸다. afti_num_tiles_in_atlas_frame_minus1의 값은 0에서 NumPartitionsInAtlasFrame-1까지의 범위에 있을 수 있다. 존재하지 않고 afti_single_partition_per_tile_flag가 1과 같을 때, afti_num_tiles_in_atlas_frame_minus1의 값은 NumPartitionsInAtlasFrame-1과 같은 것으로 추론된다.
탑 레프트 파티션 인덱스(afti_top_left_partition_idx[ i ]): i 번째 타일의 왼쪽 상단 모서리에 위치한 타일 파티션의 파티션 인덱스를 나타낸다. afti_top_left_partition_idx[ i ]의 값은 0에서 NumPartitionsInAtlasFrame-1의 범위에 있을 수 있다. 존재하지 않는 경우 afti_top_left_partition_idx[ i ]의 값은 i와 동일한 것으로 유추된다. afti_top_left_partition_idx[ i ] 신택스 요소의 길이는 Ceil( Log2( NumPartitionsInAtlasFrame ) 비트이다.
바닥 우측 파티션 컬럼 오프셋(afti_bottom_right_partition_column_offset[ i ]): i 번째 타일의 오른쪽 하단 모서리에 위치한 타일 파티션의 열 위치와 파티션 인덱스가 afti_top_left_partition_idx[ i ]와 같은 타일 파티션의 열 위치 사이의 오프셋을 나타낸다. afti_single_partition_per_tile_flag가 1과 같을 때, afti_bottom_right_partition_column_offset[ i ]의 값은 0과 같은 것으로 추론된다.
바닥 우측 파티션 로우 오프셋(afti_bottom_right_partition_row_offset[ i ]): i 번째 타일의 오른쪽 하단 모서리에 위치한 타일 파티션의 열 위치와 파티션 인덱스가 afti_top_left_partition_idx[ i ]와 같은 타일 파티션의 열 위치 사이의 오프셋을 나타낸다. afti_single_partition_per_tile_flag가 1과 같을 때, afti_bottom_right_partition_column_offset[ i ]의 값은 0과 같은 것으로 추론된다.
타일의 왼쪽 상단 및 오른쪽 하단 타일에 해당하는 타일 열과 행 위치를 지정하는 변수 topLeftColumn[ i ], topLeftRow[ i ], bottomRightColumn[ i ] 및 bottomRightRow[ i ]는 다음과 같이 계산될 수 있다:
topLeftColumn[ i ] = afti_top_left_partition_idx[ i ] % NumPartitionColumns
topLeftRow[ i ] = afti_top_left_partition_idx[ i ] / NumPartitionColumns
bottomRightColumn[ i ] = topLeftColumn[ i ] + afti_bottom_right_partition_column_offset[ i ]
bottomRightRow[ i ] = topLeftRow[ i ] + afti_bottom_right_partition_row_offset[ i ]
bottomRightColumn[ i ] 및 bottomRightRow[ i ] 값이 각각 ( asps_frame_width + 63 ) / 64-1 및 ( asps_frame_height + 63 ) / 64-1 보다 작거나 같아야 하는 것이 비트스트림 적합성의 요구 사항일 수 있다.
다음 속성 중 하나를 충족하는 j != i인 값이 없어야 하는 것도 비트스트림 적합성의 요구 사항일 수 있다.
topLeftColumn[ i ] <= topLeftColumn[ j ] <= bottomRightColumn[ i ]
topLeftRow[ i ] <= topLeftRow[ j ] <= bottomRightRow[ i ]
타일의 수평 위치, 수직 위치, 너비 및 높이를 각각 지정하는 TileOffsetX[ i ], TileOffsetY[ i ], TileWidth[ i ] 및 TileHeight[ i ] 변수는 PartitionWidth [ i ] 및 PartitionHeight[ j ]는 다음과 같이 계산될 수 있다:
TileOffsetX[ i ] = PartitionPosX[ topLeftColumn[ i ] ]
TileOffsetY[ i ] = PartitionPosY[ topLeftColumn[ i ] ]
TileWidth[ i ] = 0
TileHeight[ i ] = 0
for( j = topLeftColumn[ i ]; j <= bottomRightColumn[ i ]; j++) {
TileWidth[ i ] += PartitionWidth[ j ]
}
for( j = topLeftRow[ i ]; j <= bottomRightRow[ i ]; j++) {
TileHeight[ i ] += PartitionHeight[ j ]
}
어트리뷰트 사용 고정 카메라뷰 개수 플래그(afti_attribute_use_fixed_number_of_cameraview_flag): 실시예들에 따른 방법을 이용하여 패치 별 카메라 뷰포인트를 선택하고 이를 통해 Attribute 비디오를 생성하는 경우, 해당 타일 내 모든 패치에 대해 선택되는 카메라 뷰포인트의 개수가 동일하면 이 값은 1이 될 수 있다. 반면, 패치 단위로 선택되는 카메라 뷰포인트의 개수가 서로 다르게 적용되면 이 값은 0이 될 수 있다.
어트리뷰트 선택된 카레라뷰 카운트(afti_attribute_selected_cameraview_count[ i ]): 현재 atlas 프레임 내 i-번째 타일에 대해, 타일 내 모든 패치가 동일한 개수의 카메라 뷰포인트를 선택하도록 하는 경우, 그 개수의 값을 나타낸다. 만약 이 값이 7이면, 해당 타일 내에서는 실시예들에 따라 설명된 M = 7이며, 타일 내 모든 패치는 각각 7개의 카메라 뷰포인트를 선택한다.
오실러리 비디오 타일 로우 너비(afti_auxiliary_video_tile_row_width_minus1): 이 값에 1을 더하면, 64개의 정수 샘플 단위로 모든 보조 비디오 서브-비트스트림의 노미널 너비를 나타낸다. afti_auxiliary_video_tile_row_width_minus1이 존재하지 않을 때, 그 값은 -1과 같은 것으로 유추될 수 있다.
오실러리 비디오 타일 로우 높이(afti_auxiliary_video_tile_row_height[ i ]): 아틀라스의 i 번째 타일과 연관된 각 보조 비디오 서브 비트스트림에서 i 번째 수직 서브 영역의 64 정수 샘플 단위로 노미널 높이를 나타낸다. afti_auxiliary_video_tile_row_height[ i ]가 존재하지 않을 때, 그 값은 0과 같은 것으로 유추될 수 있다.
아틀라스의 i 번째 타일과 관련된 각 보조 하위 영역의 높이 AuxTileHeight[ i ]는 다음과 같이 계산될 수 있다:
AuxTileHeight[ i ] = afti_auxiliary_video_tile_row_height[ i ] * 64
아틀라스의 i 번째 타일과 관련된 각 보조 하위 영역의 수직 위치 AuxTileOffset[ i ]은 다음과 같이 계산될 수 있다:
AuxTileOffset[ 0 ] = 0
AuxTileOffset[ i ] = AuxTileOffset[ i - 1 ] + AuxTileHeight[ i - 1 ], for all i > 0.
아틀라스와 관련된 모든 보조 비디오 하위 비트스트림의 노미널 너비 AuxVideoWidthNF 및 높이 AuxVideoHeightNF는 다음과 같이 계산될 수 있다:
AuxVideoWidthNF = ( afti_auxiliary_video_tile_row_width_minus1 + 1 ) * 64
AuxVideoHeightNF=∑AuxTileHeight[ n ] (n=0~N)
where N is equal to afti_num_tiles_in_atlas_frame_minus1
시그널링되는 타일 아이디 플래그(afti_signalled_tile_id_flag): 1이면 각 타일에 대한 타일 ID가 있음을 나타낸다. 0과 동일한 afti_signalled_tile_id_flag는 타일 ID가 시그널링되지 않음을 나타낸다.
시그널링되는 타일 아이디 길이(afti_signalled_tile_id_length_minus1): 이 값에 1을 더하면, 존재하는 경우 구문 요소 afti_tile_id[ i ] 및 타일 헤더의 구문 요소 ath_id를 나타내는 데 사용되는 비트 수를 나타낸다. afti_signalled_tile_id_length_minus1의 값은 0에서 15까지의 범위에 있을 수 있다. 존재하지 않는 경우, afti_signalled_tile_id_length_minus1의 값은 Ceil( Log2( afti_num_tiles_in_atlas_frame_minus1 + 1 ) )-1과 동일한 것으로 추론된다.
타일 아이디(afti_tile_id[ i ]): i번째 타일의 타일 ID를 나타낸다. afti_tile_id[ i ] 신택스 요소의 길이는 afti_signalled_tile_id_length_minus1 + 1 비트이다. 존재하지 않을 때, afti_tile_id[ i ]의 값은 0에서 afti_num_tiles_in_atlas_frame_minus1까지의 범위에 있는 각각의 i에 대해 i와 동일한 것으로 추론될 수 있다. afti_tile_id[ i ]가 모든 i != j에 대해 afti_tile_id[ j ]와 동일하지 않아야 하는 것이 비트스트림 적합성의 요구사항일 수 있다. afti_tile_id[ i ] 신택스 요소의 길이는 afti_signalled_tile_id_length_minus1 + 1 비트이다.
변수 FirstTileID는 다음과 같이 계산될 수 있다:
FirstTileID = afti_tile_id[ 0 ]
for ( i = 1; i < afti_num_tiles_in_atlas_frame_minus1 + 1; i++ )
FirstTileID = Min(FirstTileID, afti_tile_id[ i ])
TileIDToIndex 및 TileIndexToID 배열은 각 타일과 연결된 ID의 정방향 및 역방향 매핑을 각각 제공하고 아틀라스 프레임 타일 정보 구문에서 각 타일이 지정된 방법의 순서 인덱스를 제공할 수 있다.
도36은 실시예들에 따른 아틀라스 어댑테이션 파라미터를 나타낸다.
아틀라스 어댑테이션 파라미터 세트 아이디(aaps_atlas_adaptation_parameter_set_id): 다른 구문 요소에서 참조하기 위해 아틀라스 적응 매개변수 세트를 식별할 수 있다.
어트리뷰트 사용 고정 카메라뷰 개수(aaps_attribute_use_fixed_number_of_camerview_flag): 실시예들에 따른 방법을 이용하여 패치 별 카메라 뷰포인트를 선택하고 이를 통해 Attribute 비디오를 생성하는 경우, 해당 atlas_adaptation_parameter_set_rbsp syntax를 참조하는 타일 유닛 내의 모든 패치에 대해 동일한 개수의 카메라 뷰포인트를 선택하도록 하고, 이 값을 atlas_adaptation_parameter_set_rbsp syntax에서 정의하고자 할 때 이 값은 1로 시그널링된다. 하지만 프레임 혹은 타일 혹은 패치 단위로 서로 다르게 그 개수가 적용되며 다른 syntax structure에서 그 값이 정의되는 경우 0으로 시그널링된다.
aaps_attribute_selected_camerview_count 해당 atlas_adaptation_parameter_set_rbsp syntax를 참조하는 타일 유닛 내의 모든 패치에 대해 동일한 개수의 카메라 뷰포인트를 선택하도록 하는 경우, 그 개수의 값을 시그널링할 수 있다. 만약 이 값이 7이면, 실시예들에 따라 설명된 M = 7이며, 해당 타일 내 모든 패치는 각각 7개의 카메라 뷰포인트를 선택한다.
로그 맥스 존재 플래그(aaps_log2_max_afoc_present_flag): 1이면 구문 요소 aaps_log2_max_atlas_frame_order_cnt_lsb가 atlas_adaptation_parameter_set_rbsp 구문 구조에 존재함을 나타낸다. 0과 동일한 aaps_log2_max_afoc_present_flag는 구문 요소 aaps_log2_max_atlas_frame_order_cnt_lsb가 존재하지 않음을 나타낸다.
로그 맥스 아틀라스 프레임 오더 카운트(aaps_log2_max_atlas_frame_order_cnt_lsb_minus4): 다음과 같이 프레임 순서 카운트에 대한 디코딩 프로세스에서 사용되는 변수 MaxAtlasFrmOrderCntLsb의 값을 계산할 수 있다:
MaxAtlasFrmOrderCntLsb = 2( aaps_log2_max_atlas_frame_order_cnt_lsb_minus4 + 4 )
aaps_log2_max_atlas_frame_order_cnt_lsb_minus4의 값은 0에서 12까지의 범위에 있을 수 있다. MaxAtlasFrmOrderCntLsb의 값은 CVS의 모든 아틀라스 하위 비트스트림에 대해 동일해야 하는 비트스트림 적합성이 필요할 수 있다.
확장 존재 플래그(aaps_extension_present_flag): 1이면 aaps_vpcc_extension_present_flag 및 aaps_extension_7bits 구문 요소가 atlas_adaptation_parameter_set_rbsp 구문 구조에 존재함을 나타낸다. 0과 동일한 aaps_extension_present_flag는 구문 요소 aaps_vpcc_extension_present_flag 및 aaps_extension_7bits가 존재하지 않음을 나타낸다.
VPCC확장 존재 플래그(aaps_vpcc_extension_present_flag): 1이면 aaps_vpcc_extension() 구문 구조가 atlas_adaptation_parameter_set_rbsp 구문 구조에 있음을 나타낸다. 0과 동일한 aaps_vpcc_extension_present_flag는 이 구문 구조가 존재하지 않음을 나타낸다. 존재하지 않을 때, aaps_vpcc_extension_present_flag의 값은 0과 같은 것으로 유추될 수 있다.
확장 비트(aaps_extension_7bits): 0이면 aaps_extension_data_flag 신택스 요소가 AAPS RBSP 신택스 구조에 존재하지 않음을 지정합니다. 존재하는 경우 aaps_extension_7bits는 이 버전의 이 문서를 준수하는 비트스트림에서 0과 같을 수 있다. 0이 아닌 aaps_extension_7bits의 값은 ISO/IEC에서 향후 사용을 위해 예약될 수 있다. 디코더는 aaps_extension_7bits 값을 0 이외의 값으로 허용하고 AAPS NAL 단위의 모든 aaps_extension_data_flag 구문 요소를 무시할 수 있다. 존재하지 않는 경우 aaps_extension_7bits의 값은 0과 동일한 것으로 유추될 수 있다.
확장 데이터 플래그(aaps_extension_data_flag): 어떤 값을 가질 수 있다. 그것의 존재와 가치는 이 문서의 이 버전에 지정된 프로파일에 대한 디코더 적합성에 영향을 미치지 않을 수 있다. 이 문서의 이 버전을 따르는 디코더는 모든 aaps_extension_data_flag 구문 요소를 무시할 수 있다.
도37은 실시예들에 따른 패치 정보 데이터를 나타낸다.
패치 모드에 따라서, 패치모드에 맞는 패치 정보를 도37과 같이 전달할 수 있다.
도38-39는 실시예들에 따른 패치 데이터 유닛을 나타낸다.
2D포지션X(pdu_2d_pos_x[ tileID ][ p ]): 현재 아틀라스 타일에서 패치 p에 대한 패치 경계 상자의 왼쪽 위 모서리의 x 좌표를 나타낸다. 타일 ID는 tileID와 같으며 PatchPackingBlockSize의 배수로 표현될 수 있다.
2D포지션Y(pdu_2d_pos_y[ tileID ][ p ] ): 현재 아틀라스 타일에서 패치 p에 대한 패치 경계 상자의 왼쪽 상단 모서리의 y 좌표를 나타낸다. 타일 ID는 tileID와 같으며 PatchPackingBlockSize의 배수로 표현될 수 있다.
2D사이즈X(pdu_2d_size_x_minus1[ tileID ][ p ]): 이 값에 1을 더하면, 타일 ID가 tileID와 동일한 현재 아틀라스 타일에서 인덱스 p가 있는 패치의 양자화된 너비 값을 나타낸다.
2D사이즈Y(pdu_2d_size_y_minus1[ tileID ][ p ]): 이 값에 1을 더하면 타일 ID가 tileID와 동일한 현재 아틀라스 타일에서 인덱스 p가 있는 패치의 양자화된 높이 값을 나타낸다.
3D오프셋Y(pdu_3d_offset_u[ tileID ][ p ]): 접선 축을 따라 타일 ID가 tileID와 동일한 현재 아틀라스 타일의 인덱스 p가 있는 패치에서 재구성된 패치 포인트에 적용할 이동을 나타낸다. pdu_3d_offset_u[ tileID ][ p ]의 값은 0에서 2^(asps_geometry_3d_bit_depth_minus1+1)-1(포함)의 범위에 있을 수 있다. pdu_3d_offset_u[ tileID ][ p ]를 나타내는 데 사용되는 비트 수는 asps_geometry_3d_bit_depth_minus1 + 1일 수 있다.
3D오프셋V(pdu_3d_offset_v[ tileID ][ p ]): 양방향 접선 축을 따라 타일 ID가 tileID와 동일한 현재 아틀라스 타일의 인덱스 p가 있는 패치에서 재구성된 패치 포인트에 적용할 이동을 나타낸다. pdu_3d_offset_v[ tileID ][ p ]의 값은 0에서 2^(asps_geometry_3d_bit_depth_minus1+1)-1(포함)의 범위에 있을 수 있다. pdu_3d_offset_v[ tileID ][ p ]를 나타내는 데 사용되는 비트 수는 asps_geometry_3d_bit_depth_minus1 + 1이다.
3D오프셋D(pdu_3d_offset_d[ tileID ][ p ]): 현재 아틀라스 타일의 인덱스 p가 있는 패치에서 재구성된 패치 포인트에 적용할 시프트를 나타낸다. 타일 ID는 tileID와 동일하며 법선 축 Pdu3dOffsetD[ tileID ][ p ]를 따라 다음과 같다: Pdu3dOffsetD[ tileID ][ p ] = pdu_3d_offset_d[ tileID ][ p ] << ath_pos_min_d_quantizer
Pdu3dOffsetD[ tileID ][ p ]의 값은 0에서 2^(asps_geometry_3d_bit_depth_minus1+1)-1(포함)의 범위에 있을 수 있다.
pdu_3d_offset_d[ tileID ][ p ]를 나타내는 데 사용되는 비트 수는 (asps_geometry_3d_bit_depth_minus1 - ath_pos_min_d_quantizer + 1)과 같을 수 있다.
3D범위D(pdu_3d_range_d[ tileID ][ p ]): 이 값이 존재하는 경우, 명목 표현으로 변환된 후 재구성된 비트 깊이 패치 기하학 샘플에 존재할 것으로 예상되는 이동의 명목상 최대값을 지정합니다. 현재 아틀라스 타일의 인덱스 p가 있는 패치에서 타일 ID가 tileID와 동일 법선 축 Pdu3dRangeD[ tileID ][ p ]는 다음과 같습니다.
if( pdu_3d_range_d[ tileID ][ p ] == 0 )
Pdu3dRangeD[ tileID ][ p ] = 0
else {
range = pdu_3d_range_d[ tileID ][ p ] << ath_pos_delta_max_d_quantizer
Pdu3dRangeD[ tileID ][ p ] = range - 1
}
변수 rangeDBitDepth를 Min(asps_geometry_2d_bit_depth_minus1, asps_geometry_3d_bit_depth_minus1) + 1과 같게 설정할 수 있다. pdu_3d_range_d[ tileID ][ p ]가 존재하지 않는 경우 Pdu3dRangeD[ ]-[ tileID ][ pDB 의 값은 Pdu3dRangeD[ ]-[pDB]로 가정될 수 있다. 존재하는 경우 Pdu3dRangeD[ tileID ][ p ] 값은 0에서 2^(rangeDBitDepth)-1(포함) 범위에 있을 수 있다.
pdu_3d_range_d[ tileID ][ p ]를 나타내는 데 사용되는 비트 수는 ( rangeDBitDepth - ath_pos_delta_max_d_quantizer )와 같을 수 있다.
pdu_projection_id[ tileID ][ p ]는 타일 ID가 tileID와 동일한 현재 아틀라스 타일의 인덱스 p를 갖는 패치에 대한 투영 평면에 대한 법선 인덱스 및 투영 모드의 값을 지정합니다. pdu_projection_id[ tileID ][ p ]의 값은 0에서 asps_max_number_projections_minus1까지의 범위에 있어야 합니다.
pdu_projection_id[ tileID ][ p ]를 나타내는 데 사용되는 비트 수는 Ceil( Log2( asps_max_number_projections_minus1 + 1) )일 수 있다.
오리엔테이션 인덱스(pdu_orientation_index[ tileID ][ p ]): tileID와 동일한 타일 ID를 가진 현재 아틀라스 타일의 인덱스 p가 있는 패치에 대해 패치 방향 인덱스를 지정하며, 이는 아틀라스를 변환하는 데 사용되는 표 11에 표시된 대로 변환 행렬 Ro 및 Rs를 결정하는 데 사용된다. 3D 공간 좌표로 변환하기 전에 좌표(u,v)로 표시되는 로컬 패치 좌표계에 대한 패치의 좌표이다. pdu_orientation_index[ tileID ][ p ]를 나타내는 데 사용되는 비트 수는 ( asps_use_eight_orientations_flag? 3 : 1 )이다.
어트리뷰트 선택된 카메라뷰 카운트(pdu_attribute_selected_cameraview_count[ tileID ][ p ]): asps_attribute_use_fixed_number_of_cameraview_flag, afps_attribute_use_fixed_number_of_cameraview_flag, aps_attribute_use_fixed_number_of_camerview_flag, afti_attribute_use_fixed_number_of_cameraview_flag 가 모두 0으로 시그널링된 것은 패치 별로 선택되는 카메라 뷰포인트의 개수가 각 패치마다 다르게 결정된다는 것을 의미한다. 이 경우, 이 syntax를 통해 타일 ID가 tileID인 현재 atlas 타일의 p번째 패치에 대해 선택된 카메라 뷰포인트의 개수를 시그널링한다.
어트리뷰트 선택된 카메라뷰(pdu_attr_selected_cameraview [ j ] 현재 패치에 대해, 선택된 카메라 뷰포인트들의 개수대로 각 카메라 뷰포인트의 정보를 시그널링한다. 이 syntax는 j번째 카메라 뷰포인트의 정보를 나타내며, 이 값은 카메라 뷰포인트의 Index가 될 수도 있고, 또는 다른 구별 가능한 인자가 될 수도 있다.
LOD인에이블 플래그(pdu_lod_enabled_flag[ tileID ][ p ]): tileID와 동일한 타일 ID를 가진 현재 아틀라스 타일의 인덱스 p가 있는 패치에 대해 패치 방향 인덱스를 지정하며, 이는 아틀라스를 변환하는 데 사용되는 표 11에 표시된 대로 변환 행렬 Ro 및 Rs를 결정하는 데 사용된다. 3D 공간 좌표로 변환하기 전에 좌표(u,v)로 표시되는 로컬 패치 좌표계에 대한 패치의 좌표입니다. pdu_orientation_index[ tileID ][ p ]를 나타내는 데 사용되는 비트 수는 ( asps_use_eight_orientations_flag? 3 : 1 )이다.
LOD스케일X(pdu_lod_scale_x_minus1[ tileID ][ p ]): 패치 좌표 TilePatch3dOffsetU[ tileID ][ p ]에 타일 ID를 추가하기 전에 타일 ID가 tileID인 현재 아틀라스 타일의 인덱스 p가 있는 패치에서 점의 로컬 x 좌표에 적용할 LOD 배율 인수를 나타낸다. pdu_lod_scale_x_minus1[ tileID ][ p ]가 존재하지 않는 경우, 그 값은 0과 동일한 것으로 유추될 수 있다.
LOD스케일Y(pdu_lod_scale_y_idc[ tileID ][ p ]): 패치 좌표 TilePatch3dOffsetV[ tileID ][ p ]에 타일 ID를 추가하기 전에 타일 ID가 tileID와 같은 현재 아틀라스 타일의 인덱스 p가 있는 패치에서 점의 로컬 y 좌표에 적용할 LOD 스케일링 계수를 나타낸다. pdu_lod_scale_y_idc[ tileID ][ p ]가 존재하지 않는 경우, 그 값은 0과 동일한 것으로 유추될 수 있다.
도40은 실시예들에 따른 V-PCC 포인트 클라우드 데이터 송신 장치를 나타낸다.
도40의 장치는 실시예들에 따른 포인트 클라우드 데이터 송신 방법/장치는 도1 송신 장치(10000), 포인트 클라우드 비디오 인코더(10002), 도4 인코딩 프로세스, 도15 비디오/이미지 인코더, 도18 송신 장치, 도20 XR디바이스(1730) 등에 대응할 수 있다. 송신 방법/장치의 각 구성요소는 하드웨어, 소프트웨어, 메모리와 연결된 프로세서, 및/또는 그것들의 조합에 대응할 수 있다.
SLF 시퀀스 데이터가 입력되면 도40의 송신기는 V-PCC에서의 동작과 같이 패치 생성부를 거쳐 입력 포인트 클라우드에 대한 패치를 생성한다(400000). 생성된 패치들에 대하여, 제안하는 패치 별 카메라 뷰포인트 선택부를 거쳐 각 패치에 대해 가장 유의미한 컬러 정보를 담고 있는 카메라 뷰포인트를 선택한다(400001). 뷰포인트 선택부(400001)는 실시예들에 따라 패치 별 카메라 뷰포인트 선택 방법 1 또는 패치 별 카메라 뷰포인트 선택 방법 2를 선택하여 뷰포인트를 선택할 수 있다. 입력된 포인트 클라우드에 대해서 생성되는 지오메트리 이미지와 오큐판시 맵은 V-PCC에서의 방법과 동일할 수 있다.
지오메트리 이미지 생성부(400003)는 포인트 클라우드 데이터의 지오메트리 데이터로부터 지오메트리 이미지를 생성한다.
패치 패킹부(400002)는 각 생성된 패치들을 2D 이미지 상으로 패킹한다. 패치 패킹된 결과를 바탕으로 제안하는 텍스처 비디오가 생성되는데, 이때 패치 별로 선택된 카메라 뷰포인트 정보를 반영하여 단일 텍스처 비디오가 생성되거나 또는 다수의 텍스처 비디오가 패치별 카레라 뷰포인트 기반 텍스쳐 이미지 생성부(400004)에 의해 생성될 수 있다.
다수의 텍스처 비디오가 생성되는 경우 생성된 비디오의 개수는 가장 많은 개수의 카메라 뷰포인트가 선택된 경우를 따라 결정된다. 텍스쳐 이미지 생성부(400004)는 단일 텍스처 비디오가 생성될지 다수의 텍스처 비디오가 생성될지 여부를 결정하고, 장치/사용자가 실시예들에 따른 Texture 비디오 생성 방법 1 과 Texture 비디오 생성 방법2 중 선택하여 텍스쳐 비디오를 생성할 수 있다. 이렇게 생성된 텍스처 비디오는 지오메트리 비디오, 오큐판시 맵과 함께 각각 2D 비디오 코덱을 이용해 압축하고 전송한다. 텍스처 비디오 생성 시 사용된 선택 방법들은 부가 정보로 시그널링되어 함께 전송된다.
부호화 전처리부는 지오메트리 이미지를 수신하고, 다수 또는 단일 텍스쳐 이미지(어트리뷰트)를 수신하여 부호화에 필요한 전처리를 수행할 수 있다.
비디오 부호화부는 지오메트리 데이터 및/또는 어트리뷰트 데이터를 부호화할 수 있다. 부보화된 지오메트리 데이터는 지오메트리 복원부에 의해 복원되어 어트리뷰트 부호화에 이용될 수 있다. 스무더는 복원된 지오메트리 데이터에 필터링 같은 처리를 적용하여, 덱스쳐 이미지 생성부에 전달할 수 있다.
메타데이터 부화화는 지오메트리 데이터(지오메트리 이미지), 어트리뷰트 데이터(텍스쳐 이미지), 오큐판시 맵에 관련된 부가 패치 정보를 부호화할 수 있다.
부가 패치 정보는 뷰포인트 선택부(400001)에 의해 생성될 수 있다.
다중화부는 인코딩된 지오메트리 데이터, 어트리뷰트 데이터, 어큐판시 맵, 부가 패치 정보를 비트스트림으로 다중화할 수 있다.
송신부는 인코딩된 포인트 클라우드 데이터를 전송할 수 있다.
도40의 송신 장치 구성요소는 도18등의 송신 장치 구성요소에 대응할 수 있다.
뷰포인트 선택부(400001)는 패치 생성부에 연결되거나, 패치 생성부에 포함될 수 있다.
텍스쳐 이미지 생성부(400004)는 패치 패킹부(400002) 및/또는 뷰포인트 선택부(400001)에 포함되거나 연결될 수 있고, 패킹 관련 전처리부(프로세서)로 송신 장치의 인코더에 연결될 수 있다.
한편, 실시예들에 따른 포인트 클라우드 데이터 수신 장치는 송신 장치의 역과정에 기반하여 포인트 클라우드 데이터를 복원할 수 있다.
수신기(도2 수신 장치(10005), 포인트 클라우드 비디오 디코더(10008), 도16 디코딩 프로세스, 도17 비디오/이미지 디코더, 도19 수신 장치, 도20 XR디바이스(1730) 등. 수신 방법/장치의 각 구성요소는 하드웨어, 소프트웨어, 메모리와 연결된 프로세서, 및/또는 그것들의 조합에 대응)로 입력된 지오메트리 비디오, 텍스처(어트리뷰트) 비디오, 오큐판시 맵 비디오와 부가 정보 데이터는 각각 디코딩되어 해당 정보를 복원하고, 이를 통해 SLF 시퀀스의 포인트 클라우드를 복원한다.
복원된 텍스처 비디오를 통해 각 포인트를 대표하는 카메라 뷰포인트의 컬러 정보를 복원하고, 이를 통해 SLF 시퀀스의 컬러 정보를 재구성하거나 목적에 맞는 컬러 정보만을 선별하여 활용할 수 있다.
복원된 부가 정보를 통해 각 비디오 별 각 패치에 해당되는 카메라 뷰포인트 정보를 알 수 있고, 이를 이용하여 포인트 클라우드 렌더링 시 현재 필요한 카메라 뷰포인트 조건에 따라 해당하는 복원된 컬러 정보를 적용할 수 있다. 만약, 현재 필요한 카메라 뷰포인트와 타겟 포인트가 갖고 있는 카메라 뷰포인트가 매칭되지 않으면, 필요한 카메라 뷰포인트와 가장 가까이에 위치한 보유 카메라 뷰포인트의 컬러 정보를 사용할 수 있다.
이로 인하여, 사용자가 상황에 따라 다수의 텍스처 비디오 비트스트림 중 중요도가 높은 일부의 텍스처 비디오 비트스트림 만을 선택하여 디코딩할 수 있는 효과가 있다. 또는 여러 개의 텍스처 비디오가 합성되어 단일 텍스처 비디오로 전송된 경우에는 필요한 텍스처 비디오에 해당하는 영역의 비트스트림 부분만을 추출하여 디코딩하고 활용할 수 있는 효과가 있다. 이를 위해서는 송신단에서 텍스처 비디오의 부분 영역에 대한 독립적 디코딩이 가능한 기능 (예, HEVC의 MCTS 기능)이 적용되어 압축될 수 있다.
V-PCC를 사용하여 SLF 데이터를 압축하는 기존의 방법 [2]은 인코딩 과정에서 발생하는 다수의 Attribute 비디오 데이터를 압축하기 위해 여러 개의 비디오 코덱 인스턴스와 높은 메모리 사용량을 필요로 하는 등의 문제를 야기한다. 이러한 문제를 개선하기 위해 제안된 다른 방법 [3]의 경우에는 지나친 데이터 축소로 인해 오히려 화질 성능을 크게 떨어트리는 부작용이 발생하게 되었다.
실시예들에 따른 압축/복원 방식은 SLF 시퀀스 압축 방식의 문제를 개선하고 보다 효율적인 압축 성능을 얻을 수 있는 효과가 있다. 패치로 구분된 포인트 클라우드 그룹에 대하여 유의미한 카메라 뷰포인트들만 선택하고, 해당 카메라 뷰포인트로부터 획득된 컬러 정보만을 전송하게함으로써 화질과 압축 성능의 효율을 향상시킨다. 패치 단위로 유의미한 카메라 뷰포인트를 선택하는 두 가지 방법들을 제안함으로써, 사용자의 활용 목적에 따라 보다 적합한 컬러 정보를 전송, 복원, 활용할 수 있도록 한다.
또한 패치 단위로 선택된 카메라 뷰포인트 정보를 이용하여 Attribute 비디오를 생성하는 방법에 대해 제안한다. 이는 모든 컬러 정보를 전송하지 않기 때문에 보다 높은 압축률로 SLF 데이터를 압축할 수 있다. 나아가 사용자의 선택에 따라 생성된 Attribute 비디오를 한 개의 비디오로 합성하여 전송함으로써 요구되는 코덱 인스턴스의 개수를 최소화하여 코딩 복잡도를 감소시킬 수 있도록 한다.
송신단과 수신단에서 수행하는 방법에 대한 정보를 확인할 수 있도록 선택된 카메라 뷰포인트 정보, 전송되는 Attribute 비디오 정보를 시그널링한다. 이로써 수신단에서는 사용자의 목적에 따라 필요한 Attribute 비디오만을 선택적으로 복원하거나 또는 모든 복원된 컬러 정보를 바탕으로 렌더링하고자 하는 방향에 적합한 컬러 정보를 사용할 수 있다.
따라서, 코딩 복잡도, 압축률, 컬러 화질, 컬러 정보 활용 관점에서 V-PCC 기반 SLF 시퀀스 데이터 압축 방법 대비 향상된 성능을 제공할 수 있다.
본 문서에서 설명하는 실시예들에 따른 동작은 실시예들에 따라서 메모리 및/또는 프로세서를 포함하는 송수신 장치에 의해 수행될 수 있다. 메모리는 실시예들에 따른 동작을 처리/제어하기 위한 프로그램들(flow chart 등) 을 저장할 수 있고, 프로세서는 본 문서에서 설명한 다양한 동작을 제어할 수 있다. 프로세서는 컨트롤러 등으로 지칭가능하다. 실시예들에 동작들은 펌웨어, 소프트웨어, 및/또는 그것들의 조합에 의해 수행될 수 있고, 펌웨어, 소프트웨어, 및/또는 그것들의 조합은 프로세서에 저장되거나 메모리에 저장될 수 있다.
실시예들에 따른 방법/장치는 V-PCC를 이용하여 일반적인 포인트 클라우드 데이터 셋 보다 많은 Attribute 속성을 갖는 Surface Light Field (SLF) 데이터 셋을 압축할 때, SLF 데이터 셋에 포함된 오브젝트 별로 유효한 카메라 뷰포인트 정보를 전송함으로써 수신단에서의 오브젝트 별 포인트 클라우드 복원 성능을 향상시키고자 한다.
다중 오브젝트를 포함하는 SLF 시퀀스를 V-PCC 로 압축하는 경우에 대해, 오브젝트 별 포인트 클라우드 복원 성능을 향상키시고, 수신단에서의 선택적인 카메라 뷰포인트 활용이 가능하게 하는 방법을 제공한다. 이것은 V-PCC 내부에서 SLF 시퀀스 데이터를 구성하는 카메라 뷰포인트들 중 오브젝트 별 유의미한 카메라 뷰포인트를 SEI message로 시그널링 (Signaling)하는 방법과 관련되어 있다.
실시예들에 따른 방법/장치는 다음 문제점을 해결하고자 한다. SLF 시퀀스 내에 하나의 오브젝트가 존재하는 것을 기본 가정할 때, 주로 압축 방식에 초점을 맞추는 제한을 제거하고자 한다.
SLF 시퀀스 내에 다수의 오브젝트가 존재하는 경우를 고려한다. 이 경우, 어떤 오브젝트는 인접한 다른 오브젝트로 인해 가려져 Attribute 정보 획득이 불가능한 카메라 뷰포인트가 발생할 수 있다. 오브젝트에 대한 Attribute 정보 획득을 하지 못한 카메라 뷰포인트는 다른 카메라 뷰포인트들로부터 획득된 Attribute 정보를 이용하여 값을 예측하고, 이 값을 자신의 Attribute 정보로 전송하게 된다. 하지만 이렇게 예측된, 다른 표현으로는 가공된 Attribute 정보가 수신단에서 다른 뷰포인트의 Attribute 값을 복원하는데 사용된다면, 이는 복원된 포인트 클라우드의 화질 열화에 영향을 줄 수 있다. 그러므로 본 발명에서는 가공된 정보는 배제하고 정확한 정보만을 이용하여 Attribute을 복원함으로써 불필요한 복원 화질 열화를 막는 효과를 제공할 수 있다.
실시예들에 따른 방법/장치가 V-PCC를 이용하여 다중 오브젝트를 포함하는 SLF 시퀀스를 코딩할 때, 오브젝트 별 사용가능한 카메라 뷰포인트 정보를 전송하는 방법 및 관련 시그널링 방법을 제공한다.
일괄적인 카메라 뷰포인트 정보가 아닌 오브젝트 별 유의미한 카메라 뷰포인트를 이용하게 함으로써 복원 화질의 성능 열화를 최소화할 수 있는 효과가 있다.
오브젝트 별 유효한 카메라 뷰포인트 정보를 바탕으로, 수신단에서 선택적인 카메라 뷰포인트의 활용이 용이한 효과가 있다.
도41은 실시예들에 따른 다중 오브젝트를 포함하는 SLF 데이터 셋 예시를 나타낸다.
실시예들에 따른 방법/장치는 오브젝트#1 및 오브젝트#2와 같이 다중 오브젝트에 관한 포인트 클라우드 데이터를 압축하고 복원할 수 있다.
도41의 화살표는 서로 다른 멀티플 카메라 뷰인트들을 나타낸다.
도41에서 보이는 바와 같이, SLF 데이터 셋은 다수의 카메라 뷰포인트들로부터 획득된 각 포인트에 대한 Attribute 정보를 모두 데이터로 포함하며, V-PCC는 이와 같이 생성된 SLF 데이터 셋을 압축, 전송한다. 만약 SLF 데이터 셋이 다수의 오브젝트를 포함하고 있는 경우에는, 도41예시와 같이 인접한 다른 오브젝트로 인해 가려져 Attribute 정보 획득이 불가능한 카메라 뷰포인트가 발생할 수 있다.
도41의 오른쪽 오브젝트 object #2는 왼쪽 오브젝트 object #1로 인해 가려져 정보 획득이 불가능한 두 개의 카메라 뷰포인트가 발생하고(41000), 이로 인해 object #2의 가려진 영역(Occluded area)으로 표시된 영역에 포함되는 포인트들은 이 두 개의 카메라 뷰포인트를 통한 Attribute 정보 획득이 불가능하게 된다.
마찬가지로, 오브젝트1은 오브젝트 2에 의해 가려져 획득하지 못하는 카레라 뷰포인트 영역(41001)이 발생한다.
하지만 SLF 데이터 셋을 생성할 때에는 이와 같이 획득되지 못한 Attribute 정보는 다른 사용가능한 카메라 뷰포인트로부터 획득된 Attribute 정보를 이용하여 가공된 값으로 대체되어 생성, 사용될 것이다. 이렇게 생성된 데이터 셋은 V-PCC를 통해 전송되어 포인트 클라우드 복원에 사용된다. 가공된 Attribute 정보는 획득되지 못한 정보를 대체할 수는 있지만, 이렇게 가공된 정보를 활용하여 다른 주변 포인트들의 Attribute 값을 복원할 경우에는 데이터의 정확도를 떨어트리는 요인이 될 수 있으며, 이로 인해 복원 화질의 열화에 영향을 미칠 수 있다.
본 발명에서는 이와 같은 불필요한 화질 열화를 최소화하기 위하여, 포인트의 Attribute 정보 복원 시 가공되지 않은, 직접 획득된 데이터만을 활용할 수 있도록 오브젝트 별로 사용가능한 카메라 뷰포인트의 정보를 추가로 시그널링하여 전송하는 방법을 제안한다. 제안하는 방법에서는 SLF 시퀀스에 포함된 모든 오브젝트들에 대해, 각 오브젝트 별로 사용할 수 있는 카메라 뷰포인트 정보를 SEI message 에 포함하여 전송하는 방법을 제안한다. SEI message에 포함되는 정보는 오브젝트 별 유효한 카메라 뷰포인트의 개수와 각 카메라 뷰포인트를 식별할 수 있는 인덱스 정보이다.
실시예들에 따른 가려진 영역 관련 정보들은 V-PCC의 Volumetric annotation SEI message family 중 Scene object information SEI message 에 포함되어 시그널링된다(이하, 도42-47).
실시예들에 따른 포인트 클라우드 데이터 송신 장치는 도42-47과 같은 정보를 생성하고, 비트스트림에 포함시켜서, 수신 장치에 전송할 수 있다.
실시예들에 따른 포인트 클라우드 데이터 수신 장치는 포인트 클라우드 데이터를 포함하는 비트스트림을 수신하고, 비트스트림에 포함된 도42-47정보에 기반하여 비트스트림에 포함된 포인트 클라우드 데이터(지오메트리 데이터, 어트리뷰트 데이터, 어큐판시 맵 등)을 디코딩할 수 있다.
도42-47은 실시예들에 따른 볼륨메트릭 어노테이션 SEI메시지(Volumetric annotation SEI message family syntax)에 포함되는 신 오브젝트 정보 SEI메시지 신택스(Scene object information SEI message syntax )를 나타낸다.
이 SEI 메시지는 체적 장면에 존재할 수 있는 객체 세트를 정의하고 선택적으로 이러한 객체에 다른 속성을 할당한다. 그런 다음 이러한 객체는 패치 정보 및 체적 직사각형 정보 SEI 메시지를 사용하여 정의될 수 있는 패치 및 2D 체적 직사각형을 포함하여 다른 유형의 정보와 잠재적으로 연관될 수 있다.
각 시퀀스의 시작 부분에서 ObjectTracked[ k ] = 0으로 설정한다. 여기서 k는 객체 인덱스에 해당하고 0에서 232 -1까지의 범위에 있다. ObjectTracked[ k ]가 0이면 개체 레이블, 3D 경계 상자 매개 변수, 우선 순위 정보, 숨겨진 플래그, 종속성 정보, 가시 원뿔, 충돌 모양, 점 스타일 및 재료 ID를 포함한 모든 관련 매개변수가 기본 값을 포함함을 나타낸다.
필요한 메모리를 제한하기 위해 응용 프로그램에서 개체 인덱스 제한을 추가로 지정할 수 있다.
지속성 플래그(soi_persistence_flag): 현재 레이어에 대한 장면 객체 정보 SEI 메시지의 지속성을 나타낸다. 0과 동일한 soi_persistence_flag는 장면 객체 정보 SEI 메시지가 현재 디코딩된 아틀라스 프레임에만 적용되는 것을 나타낸다.
aFrmA를 현재 아틀라스 프레임이라고 한다면, 1과 동일한 soi_persistence_flag는 장면 객체 정보 SEI 메시지가 다음 조건 중 하나가 참일 때까지 출력 순서로 현재 레이어에 대해 지속되는 것을 나타낼 수 있다:
1. 새로운 CAS가 시작
2. 비트스트림 끝
3. Soi_persistence_flag의 동일한 값을 갖는 장면 객체 정보 SEI 메시지를 포함하고 현재 레이어에 적용 가능한 코딩된 아틀라스 액세스 단위의 현재 레이어에 있는 아틀라스 프레임 aFrmB가 출력되며, 여기서 AtlasFrmOrderCnt( aFrmB )가 AtlasFrmOrderCnt( aFrmA )보다 더 크다. AtlasFrmOrderCnt( aFrmB ) 및 AtlasFrmOrderCnt( aFrmA )는 aFrmB에 대한 아틀라스 프레임 순서 카운트에 대한 디코딩 프로세스 호출 직후의 aFrmB 및 aFrmA의 AtlasFrmOrderCntVal 값이다.
리셋 플래그(soi_reset_flag): 이 장면 객체 정보 SEI 메시지에 해당하는 정보가 기본값으로 재설정되었음을 나타낸다.
오브젝트 개수 업데이트(soi_num_object_updates): 현재 SEI에 의해 업데이트될 객체의 수를 나타낸다. soi_num_object_updates의 값은 0에서 232-1까지의 범위에 있을 수 있다. soi_num_object_updates의 기본값은 0이다.
심플 오브젝트 플래그(soi_simple_objects_flag): 1과 같으면 업데이트되거나 새로 도입된 객체에 대한 추가 정보가 신호되지 않음을 나타낸다. 0과 동일한 soi_simple_objects_flag는 업데이트되거나 새로 도입된 객체에 대한 추가 정보가 시그널링될 수 있음을 나타낸다.
오브젝트 라벨 존재 플래그(soi_object_label_present_flag): 1과 같음은 객체 레이블 정보가 현재 장면 객체 정보 SEI 메시지에 존재함을 나타낸다. 0과 동일한 soi_object_label_present_flag는 객체 레이블 정보가 존재하지 않음을 나타낸다.
우선선위 존재 플래그(soi_priority_present_flag): 1과 같음은 우선순위 정보가 현재 장면 객체 정보 SEI 메시지에 존재함을 나타낸다. 0과 동일한 soi_priority_present_flag는 우선순위 정보가 존재하지 않음을 나타낸다.
오브젝트 히든 존재 플래그(soi_object_hidden_present_flag): 1과 같음은 숨겨진 객체 정보가 현재 장면 객체 정보 SEI 메시지에 존재함을 나타낸다. 0과 동일한 soi_object_hidden_present_flag는 히든 오브젝트 정보가 존재하지 않음을 나타낸다.
오브젝트 디펜던시 존재 플래그(soi_object_dependency_present_flag): 1과 같음은 객체 종속성 정보가 현재 장면 객체 정보 SEI 메시지에 존재함을 나타낸다. 0과 동일한 soi_object_dependency_present_flag는 객체 의존성 정보가 존재하지 않음을 나타낸다.
가시성 콘 존재 플래그(soi_visibility_cones_present_flag): 1과 같음은 가시성 원뿔 정보가 현재 장면 객체 정보 SEI 메시지에 존재함을 나타낸다. 0과 동일한 soi_visibility_cones_present_flag는 가시성 원뿔 정보가 존재하지 않음을 나타낸다.
3D 바운딩 박스 존재 플래그(soi_3d_bounding_box_present_flag): 1과 같음은 3D 경계 상자 정보가 현재 장면 객체 정보 SEI 메시지에 존재함을 나타낸다. 0과 동일한 soi_3d_bounding_box_present_flag는 3D 경계 상자 정보가 존재하지 않음을 표시한다.
충돌 모양 존재 플래그(soi_collision_shape_present_flag): 1과 같음은 충돌 정보가 현재 장면 객체 정보 SEI 메시지에 존재함을 나타낸다. 0과 동일한 soi_collision_shape_present_flag는 충돌 형태 정보가 존재하지 않음을 나타낸다.
포인트 스타일 존재 플래그(soi_point_style_present_flag): 1과 같음은 포인트 스타일 정보가 현재 장면 객체 정보 SEI 메시지에 존재함을 나타낸다. 0과 동일한 soi_point_style_present_flag는 포인트 스타일 정보가 존재하지 않음을 나타낸다.
머터리얼 ID존재 플래그(soi_material_id_present_flag): 1과 같음은 재료 ID 정보가 현재 장면 객체 정보 SEI 메시지에 존재함을 나타낸다. 0과 동일한 soi_material_id_present_flag는 재료 ID 정보가 존재하지 않음을 나타낸다.
확장 존재 플래그(soi_extension_present_flag): 1과 같음은 추가 확장 정보가 현재 장면 객체 정보 SEI 메시지에 존재해야 함을 나타낸다. 0과 동일한 soi_extension_present_flag는 추가 확장 정보가 존재하지 않음을 나타낸다. soi_extension_present_flag가 0과 같아야 하는 것은 이 문서의 이 버전에 대한 비트스트림 적합성의 요구사항이다.
오브젝트 카메라뷰 존재 플래그(soi_object_cameraview_present_flag): 이 값이 1이면 object 별 카메라 뷰포인트 정보가 현재의 scene object information SEI message에 존재함을 의미한다. 이 값이 0이면 해당 정보가 존재하지 않는다.
3D바운딩 박스 스케일로그(soi_3d_bounding_box_scale_log2): 개체에 대해 지정할 수 있는 3D 경계 상자 매개변수에 적용할 배율을 나타낸다.
3D바운딩 박스 프리시전(soi_3d_bounding_box_precision_minus8): 이 값에 8을 더하면, 개체에 대해 지정할 수 있는 3D 경계 상자 매개변수의 정밀도를 나타낸다.
로그맥스 오브젝트 인덱스 업데이트(soi_log2_max_object_idx_updated_minus1): 이 값에 1을 더하면, 현재 장면 객체 정보 SEI 메시지에서 객체 인덱스 값을 시그널링하는 데 사용되는 비트 수를 나타낸다.
로그맥스 오브젝트 디펜던시 인덱스(soi_log2_max_object_dependency_idx): 현재 장면 객체 정보 SEI 메시지에서 종속성 객체 인덱스 값을 시그널링하는 데 사용되는 비트 수를 나타낸다. soi_log2_max_object_dependency_idx의 기본값은 0입니다.
오브젝트 인덱스(soi_object_idx[ i ]): 업데이트할 i번째 객체의 객체 인덱스를 나타낸다. soi_object_idx[ i ]를 나타내는 데 사용되는 비트 수는 soi_log2_max_object_idx_updated_minus1 + 1과 같다. 비트스트림에 soi_object_idx[ i ]가 없으면 그 값은 0과 동일한 것으로 유추된다.
오브젝트 캔슬 플래그(soi_object_cancel_flag[ i ]): 1과 같음은 인덱스가 i인 객체가 취소되고 ObjectTracked[ i ] 변수가 0으로 설정됨을 나타낸다. 또한 객체 레이블, 3D 경계 상자 매개변수, 우선순위 정보, 숨겨진 플래그, 종속성 정보, 가시성 원뿔, 충돌 모양, 포인트 스타일 및 재질 ID가 기본값으로 재설정된다. 0과 동일한 soi_object_cancel_flag는 인덱스가 soi_object_idx[ i ]인 객체가 이 요소를 뒤따르는 정보로 업데이트되고 변수 ObjectTracked[ i ]가 1로 설정되어야 함을 나타낸다.
오브젝트 라벨 업데이트 플래그(soi_object_label_update_flag[ i ]): 1과 같으면 객체 인덱스 i를 가진 객체에 대한 객체 레이블 업데이트 정보가 있음을 나타낸다. 0과 동일한 soi_object_label_update_flag[ i ]는 객체 레이블 업데이트 정보가 존재하지 않음을 나타낸다.
오브젝트 라벨 인덱스(soi_object_label_idx[ i ]): 인덱스 i가 있는 객체의 레이블 인덱스를 나타낸다. soi_object_label_idx[ i ]의 값은 0에서 232-1까지의 범위에 있을 수 있다.
우선순위 업데이트 플래그(soi_priority_update_flag[ i ]): 1과 같으면 객체 인덱스 i를 가진 객체에 대한 우선 순위 업데이트 정보가 있음을 나타낸다. 0과 동일한 soi_priority_update_flag[ i ]는 객체 우선순위 정보가 존재하지 않음을 나타낸다.
우선순위 값(soi_priority_value[ i ]): 인덱스 i가 있는 객체의 우선 순위를 나타낸다. 우선 순위 값이 낮을수록 우선 순위가 높아진다. soi_priority_value[ i ]의 기본값은 0이다.
오브젝트 히든 플래그(soi_object_hidden_flag[ i ]): 1과 같으면 인덱스 i가 있는 개체가 숨겨짐을 나타낸다. 0과 같은 soi_object_hidden_flag[ i ]는 인덱스 i를 가진 객체가 존재하게 될 것임을 나타낸다.
오브젝트 디펜던시 업데이트 플래그(soi_object_dependency_update_flag[ i ]): 1과 같으면 객체 인덱스 i를 가진 객체에 대한 객체 종속성 업데이트 정보가 있음을 나타낸다. 0과 동일한 soi_object_dependency_update_flag[ i ]는 객체 종속성 업데이트 정보가 존재하지 않음을 나타낸다.
오브젝트 디펜던시 개수(soi_object_num_dependencies[ i ]): 인덱스 i가 있는 객체의 종속성 수를 나타낸다.
오브젝트 디펜던시 인덱스(soi_object_dependency_idx[ i ][ j ]): 객체 인덱스 i를 가진 객체와 종속성을 갖는 j번째 객체의 인덱스를 나타낸다.
가시성 콘 업데이트 플래그(soi_visibility_cones_update_flag[ i ]): 1과 같으면 가시성 원뿔 업데이트 정보가 객체 인덱스 i를 가진 객체에 대해 존재한다는 것을 나타낸다. 0과 동일한 soi_visibility_cones_update_flag[ i ]는 가시성 원뿔 업데이트 정보가 존재하지 않음을 나타낸다.
다이렉션X(soi_direction_x[ i ]): 객체 인덱스가 i인 객체의 가시성 원뿔에 대한 방향 벡터의 정규화된 x 성분 값을 나타낸다. soi_direction_x[ i ]의 값은 존재하지 않을 때 1.0과 동일한 것으로 추론된다. soi_direction_x[ i ]의 기본값은 1.0과 같다.
다이렉션Y(soi_direction_y[ i ]): 객체 인덱스가 i인 객체의 가시성 원뿔에 대한 방향 벡터의 정규화된 y 성분 값을 나타낸다. soi_direction_y[ i ]의 값은 존재하지 않을 때 1.0과 같은 것으로 추론된다. soi_direction_y[ i ]의 기본값은 1.0과 같다.
다이렉션Z(soi_direction_z[ i ]): 객체 인덱스가 i인 객체의 가시성 원뿔에 대한 방향 벡터의 정규화된 z 성분 값을 나타낸다. soi_direction_z[ i ]의 값은 존재하지 않을 때 1.0과 동일한 것으로 추론된다. soi_direction_z[ i ]의 기본값은 1.0과 같다.
앵글(soi_angle[ i ]): 방향 벡터를 따라 가시 원뿔의 각도를 도 단위로 나타낸다. soi_angle[ i ]의 값은 존재하지 않을 때 180과 같은 것으로 추론됩니다. soi_angle[ i ]의 기본값은 180과 같다.
3D바운딩 박스 업데이트 플래그(soi_3d_bounding_box_update_flag[ i ]): 1과 같으면 객체 인덱스 i를 가진 객체에 대해 3D 경계 상자 정보가 있음을 나타낸다. 0과 동일한 soi_3d_bounding_box_update_flag[ i ]는 3D 경계 상자 정보가 존재하지 않음을 나타낸다.
3D바운딩 박스 X(soi_3d_bounding_box_x[ i ]): 인덱스 i를 가진 객체의 3D 경계 상자의 원점 위치의 x 좌표 값을 나타낸다. soi_3d_bounding_box_x[ i ]의 기본값은 0이다.
3D바운딩 박스 Y(soi_3d_bounding_box_y[ i ]): 인덱스 i를 가진 객체의 3D 경계 상자의 원점 위치의 y 좌표 값을 나타낸다. soi_3d_bounding_box_y[ i ]의 기본값은 0이다.
3D바운딩 박스Z(soi_3d_bounding_box_z[ i ]): 인덱스 i를 가진 객체의 3D 경계 상자의 원점 위치의 z 좌표 값을 나타낸다. soi_3d_bounding_box_z[ i ]의 기본값은 0이다.
3D바운딩 박스 델타X( soi_3d_bounding_box_delta_x[ i ]): 인덱스가 i인 객체의 x축에 있는 경계 상자의 크기를 나타낸다. soi_3d_bounding_box_delta_x[ i ]의 기본값은 0이다.
3D바운딩 박스 델타 Y(soi_3d_bounding_box_delta_y[ i ]): 인덱스가 i인 객체의 y축에 있는 경계 상자의 크기를 나타낸다. soi_3d_bounding_box_delta_y[ i ]의 기본값은 0이다.
3D바운딩 박스 델타 Z(soi_3d_bounding_box_delta_z[ i ]): 인덱스가 i인 객체의 z축에 있는 경계 상자의 크기를 나타낸다. soi_3d_bounding_box_delta_z[ i ]의 기본값은 0이다.
충돌 모양 업데이트 플래그(soi_collision_shape_update_flag[ i ]): 1과 같으면 객체 인덱스 i를 가진 객체에 대한 충돌 모양 업데이트 정보가 있음을 나타낸다. 0과 동일한 soi_collision_shape_update_flag[ i ]는 충돌 형상 업데이트 정보가 존재하지 않음을 표시한다.
충돌 모양 아이디(soi_collision_shape_id[ i ]): 인덱스가 i인 물체의 충돌 모양 ID를 나타낸다. 충돌 형상 ID는 이 문서 외부의 수단을 통해 식별된다. soi_collision_shape_id[ i ]의 기본값은 0이다.
포인트 스타일 업데이트 플래그(soi_point_style_update_flag[ i ]): 1과 같으면 객체 인덱스 i를 가진 객체에 대한 포인트 스타일 업데이트 정보가 있음을 나타낸다. 0과 동일한 soi_point_style_update_flag[ i ]는 포인트 스타일 업데이트 정보가 존재하지 않음을 나타낸다.
포인트 모양 아이디(soi_point_shape_id[ i ]): 인덱스 i가 있는 개체의 점 모양 ID를 나타낸다. soi_point_shape_id[ i ]의 기본값은 0과 같다. soi_point_shape_id[ i ]의 값은 이 버전의 이 문서를 준수하는 비트스트림을 포함하여 0에서 2의 범위에 있을 수 있다. soi_point_shape_id[ i ]의 다른 값은 ISO/IEC에서 향후 사용을 위해 예약되어 있을 수 있다. 이 문서의 이 버전을 따르는 디코더는 soi_point_shape_id[ i ]의 예약된 값을 무시할 수 있다.
soi_point_shape_id[ i ] 값들: 0이면, Circle(원형), 1이면, Square(사각형), 2이면, Diamond(다이아몬드), 3 내지 255이면 Reserved(예약)일 수 있다.
포인트 사이즈(soi_point_size[ i ]): 인덱스 i가 있는 객체의 포인트 크기를 나타낸다. soi_point_size[ i ]의 기본값은 1과 같다.
머터리얼 아이디 업데이트 플래그(soi_material_id_update_flag[ i ]): 1과 같으면 객체 인덱스 i를 가진 객체에 대해 재료 ID 업데이트 정보가 있음을 나타낸다. 0과 동일한 soi_point_style_update_flag[ i ]는 포인트 스타일 업데이트 정보가 존재하지 않음을 나타낸다.
머터리얼 아이디(soi_material_id[ i ]): 인덱스 i가 있는 개체의 재질 ID를 나타낸다. soi_material_id[ i ]의 기본값은 0이다. 재료 ID는 이 문서 외부의 수단을 통해 식별된다.
오브젝트 카메라뷰 업데이트 플래그(soi_object_cameraview_update_flag[ i ]): 이 값이 1이면 i번째 object에 대한 카메라 뷰포인트 update 정보가 존재함을 나타낸다. 이 값이 0이면 해당 정보가 존재하지 않음을 나타낸다.
오브젝트 카메라뷰 개수(soi_object_num_cameraviews[ i ]): i번째 object에 대해 유효한 카메라 뷰포인트들의 개수를 나타낸다.
오브젝트 카메라뷰 인덱스(soi_object_cameraview_idx[ i ][ j ]): i번째 object에 대해 유효한 카메라 뷰포인트들 중 j 번 째 카메라 뷰포인트의 index를 나타낸다. 또는 index 외에 카메라 뷰포인트를 구별할 수 있는 다른 정보의 값으로도 대체하여 사용할 수 있다.
실시예들에 따른 포인트 클라우드 데이터 송신 방법/장치는 다음과 같이 포인트 클라우드 데이터를 인코딩하고 전송할 수 있다.
SLF 데이터 셋을 생성할 때 현재의 오브젝트가 주변 오브젝트로 인해 가려지는 영역이 없는 경우, 해당 오브젝트는 존재하는 모든 카메라 뷰포인트를 통해 Attribute 획득이 가능하다고 판단하고 이에 해당하는 카메라 뷰포인트 정보를 제안하는 SEI message를 이용하여 전송한다. 하지만 주변 오브젝트로 인해 가려지는 영역이 발생하여 해당 방향으로 존재하는 카메라 뷰포인트로부터 Attribute 정보 획득이 불가능한 경우(도41 참조), 이들을 제외한 후 나머지 사용 가능한 카메라 뷰포인트들의 정보들만을 해당 오브젝트의 SEI message로 전송한다(도42-47 참조).
여기에서 카메라 뷰포인트가 가려지는 것은 동일 오브젝트에 존재하는 포인트 영역으로 인해 카메라 뷰포인트가 가려지는 상황이 아닌, 다른 오브젝트로 인해 가려지는 상황만을 고려하여 판단한다.
실시예들에 따른 포인트 클라우드 데이터 송신 방법/장치는 다음과 같이 포인트 클라우드 데이터를 수신하고 디코딩할 수 있다.
수신된 비트스트림에 포함된 SEI message를 통해, 다중 오브젝트가 포함된 SLF 시퀀스에 대하여 각 오브젝트 별 사용 가능한 카메라 뷰포인트 정보를 확인할 수 있다. 오브젝트의 구분 없이 모든 카메라 뷰포인트 정보를 사용하여 Attribute 정보를 복원하는 방법과 달리, 실시예들에 따른 방법을 통해 오브젝트 별로 사용할 수 있는 카메라 뷰포인트의 정보를 확인, 이들을 바탕으로 Attribute 데이터를 복원한다. 제안 방법에서는 데이터 복원 시 가공된 정보는 배제되기 때문에 불필요한 화질 열화를 막을 수 있다.
또한 수신된 비트스트림에 포함된 정보를 바탕으로 사용자의 의도에 따라 오브젝트 별로 선택적인 카메라 뷰포인트의 활용이 가능하다. 예를 들어 사용자가 타겟 오브젝트의 특정 방향에 대해서만 복원하고자 할 경우, 해당 방향을 포함하는 일정 영역 내에 존재하는 카메라 뷰포인트를 판단하고 수신된 SEI 정보를 통해 이들 중 실제 사용가능한 카메라 뷰포인트 정보를 확인, 렌더링에 활용할 수 있다. 또는 사용자가 수신된 정보로부터 유효한 카메라 뷰포인트들을 확인 후, 이들 중에서 보다 유의미하다고 판단되는 특정 카메라 뷰포인트들만을 선별하여 데이터 복원에 활용할 수 있다.
실시예들에 따른 방법/장치는 SLF 시퀀스 압축 기술에서 고려하지 않았던, 다중 오브젝트를 포함하는 SLF 포인트 클라우드 데이터 셋에 대한 새로운 카메라 뷰포인트 정보 처리 방법을 제공한다. 다중 오브젝트가 같은 3차원 공간에 존재하는 경우, 서로 다른 오브젝트 간의 간섭으로 인해 각 오브젝트에서는 Attribute 정보 획득이 불가능한 카메라 뷰포인트가 발생할 수 있다. 이 경우, SLF 데이터 특성 상 가공된 Attribute 정보가 전송되지만 가공된 정보가 디코더에서 다른 포인트들의 Attribute 정보를 복원하는데 사용되는 경우, 복원 화질 열화는 불가피하다. 실시예들은 이러한 문제를 해결하기 위해, 디코더에서 Attribute 복원 시 실제 카메라 뷰포인트로부터 획득된 Attribute 정보만을 사용할 수 있도록 각 오브젝트 별로 사용가능한 카메라 뷰포인트에 대한 정보를 시그널링하여 전송할 수 있다. V-PCC에서 오브젝트 별로 서로 다른 특성 정보를 전송할 수 있도록 하는 Scene object information SEI message에 새로운 syntax를 추가하여 SLF 시퀀스 내 오브젝트 별 사용할 수 있는 카메라 뷰포인트의 정보를 전송하고 수신할 수 있다. 수신 방법/장치는 다중 Attribute 정보를 이용하여 포인트 클라우드를 복원하는 경우 각 오브젝트 별로 유효한 카메라 뷰포인트들의 Attribute 정보만을 활용하여 복원하게 함으로써 불필요한 화질 열화를 막을 수 있다. 또한, 사용자의 의도에 따라 오브젝트 별 필요한 카메라 뷰포인트들을 선택적으로 사용할 수 있도록 함으로써 수신단에서의 Attribute 정보 활용 유연성과 정확도 향상 효과를 제공한다.
도48은 실시예들에 따른 포인트 클라우드 데이터 송신 방법을 나타낸다.
S4800 실시예들에 따른 포인트 클라우드 데이터 송신 방법은 포인트 클라우드 데이터를 인코딩하는 단계를 포함할 수 있다.
실시예들에 따른 인코딩 동작은 도1 송신 장치(10000), 포인트 클라우드 비디오 인코더(10002), 파일/세그먼트 인캡슐레이터(10003), 도4 인코딩 프로세서, 도15 인코더, 도18 송신 장치, 도20 XR디바이스(2030), SLT데이터를 포함하는 도21-25 오브젝트 인코딩, 도26 비트스트림 생성, 도27내지 도39 파라미터 정보 생성, 도40 송신 장치, 도41 다수 오브젝트들 인코딩, 도42 내지 도47 파라미터 정보 생성 등의 동작을 포함할 수 있다.
S4801 실시예들에 따른 포인트 클라우드 데이터 송신 방법은 포인트 클라우드 데이터를 포함하는 비트스트림을 전송하는 단계를 더 포함할 수 있다.
실시예들에 따른 전송 동작은 도1 송신 장치(10000), 트랜스미터(10004), 도4 비트스트림 전달, 도15 비트스트림 전달, 도18 송신 장치의 비트스트림 송신, 도20 XR디바이스(2030)의 비트스트림 전달, 도21 내지 도39 인코딩된 포인트 클라우드 데이터를 포함하는 비트스트림 전달, 도40 송신 장치의 포인트 클라우드 데이터 전송, 도41 내지 도47 인코딩된 포인트 클라우드 데이터를 포함하는 비트스트림 전송 등을 포함할 수 있다.
도49는 실시예들에 따른 포인트 클라우드 데이터 수신 방법을 나타낸다.
S4800 실시예들에 따른 포인트 클라우드 데이터 수신 방법은 포인트 클라우드 데이터를 포함하는 비트스트림을 수신하는 단계를 포함할 수 있다.
실시예들에 따른 수신 동작은 도1 수신 장치(10005), 리시버(10006), 파일/세그먼트 디캡슐레이터(10007), 도16-17, 19 포인트 클라우드 데이터를 포함하는 비트스트림 수신, 도20 XR 디바이스(2030) 포인트 클라우드 데이터 수신, 도21 내지 도25 데이터 셋을 포함하는 포인트 클라우드 데이터 수신, 도26 내지 도39 비트스트림에 포함된 데이터 수신, 도41 멀티플 오브젝트들에 대한 포인트 클라우드 데이터 수신, 도42 내지 도47 비트스트림에 포함된 데이터 수신 등의 동작을 포함할 수 있다.
S4801 실시예들에 따른 포인트 클라우드 데이터 수신 방법은 포인트 클라우드 데이터를 디코딩하는 단계를 더 포함할 수 있다.
실시예들에 따른 디코딩 동작은 도1 포인트 클라우드 비디오 디코더(10008), 도16-17 디코딩, 도19 수신 장치, 도20 XR디바이스(2030)의 디코딩, 도21 내지 도25 데이터 셋을 포함하는 포인트 클라우드 데이터 디코딩, 도26 내지 도39 비트스트림에 포함된 데이터 디코딩, 도41 멀티플 오브젝트들에 대한 포인트 클라우드 데이터 디코딩, 도42 내지 도47 비트스트림에 포함된 데이터 디코딩 등의 동작을 포함할 수 있다.
실시예들에 따른 포인트 클라우드 데이터 송신 방법은 포인트 클라우드 데이터를 인코딩하는 단계; 및 포인트 클라우드 데이터를 포함하는 비트스트림을 전송하는 단계; 를 포함하할 수 있다.
실시예들에 따른 포인트 클라우드 데이터는 뷰포인트들에 대한 카메라들로부터 획득된 지오메트리 데이터(오브젝트의 포인트의 위치값) 및 적어도 두 개의 어트리뷰트들(예를 들어, 복수의 컬러값들)을 포함하고, 포인트 클라우드 데이터 송신 방법은 오브젝트 및 상기 뷰포인트들 간 거리에 기반하여 상기 뷰인트들로부터 특정 개수의 뷰포인트들을 선택하는 단계를 더 포함할 수 있다.
실시예들에 따른 포인트 클라우드 데이터 송신 방법은 선택된 특정 개수의 뷰포인트들로부터 대표 어트리뷰트 정보를 생성하고, 대표 어트리뷰트 정보에 기반하여 특정 개수의 뷰포인트들을 선택하는 단계를 더 포함할 수 있다. 대표 어트리뷰트 정보(예를 들어, 특정 포인트의 대표 컬러값)과 차이가 많이 나는 어트리뷰트를 우선 압축하는 것이 효율적이기 때문이다.
또한, 거리 순서(예를 들어, 거리 오름차순 또는 거리 내림차순 등)에 기반하여, 선택된 특정 개수의 뷰포인트들에 대한 어트리뷰트 정보로부터 텍스쳐 데이터를 생성할 수 있다. 텍스쳐 데이터는 텍스쳐(어트리뷰트) 비디오 또는 텍스쳐(어트리뷰트) 이미지 등을 의미할 수 있다.
포인트 클라우드 데이터 송신 방법은 대표 어트리뷰트 정보 및 특정 개수의 뷰포인트들에 대한 어트리뷰트 정보 간 차이에 기반하여, 특정 개수의 뷰포인트들로부터 텍스쳐 데이터를 생성하는 단계를 더 포함할 수 있다.
포인트 클라우드 데이터 송신 방법은 상기 선택된 특정 개수의 뷰포인트들에 대한 어트리뷰트 정보를 포함하는 텍스쳐 데이터를 머지하여 하나의 텍스쳐 데이터를 생성하는 단계를 더 포함하는,
실시예들에 따라 생성되고 인코딩된 비트스트림은 카메라 뷰인트 선택에 관련된 파라미터 정보를 포함할 수 있다.
실시예들에 따른 포인트 클라우드 데이터는 오브젝트들로부터 획득된 지오메트리 데이터 및 적어도 두 개의 어트리뷰트들을 포함하고, 비트스트림은 제1오브젝트 및 제2오브젝트 간 가려진 영역에 기반하여 생성된 유효한 카메라 뷰포인트 관련 파라미터 정보를 포함할 수 있다. 가려진 영역은 포인트 클라우드 데이터 복원 시 불필요할 수 있기 때문에, 가려진 영역을 제외하여 비트스트림 사이즈를 줄여서 정확하고 효율적인 압축 및 복원이 가능한 효과가 있다.
실시예들에 따른 포인트 클라우드 데이터 수신 장치는 포인트 클라우드 데이터를 포함하는 비트스트림을 수신하는 리시버; 및 포인트 클라우드 데이터를 디코딩하는 디코더; 를 포함할 수 있다.
실시예들에 따른 포인트 클라우드 데이터는 뷰포인트들에 대한 카메라들로부터 획득된 지오메트리 데이터 및 적어도 두 개의 어트리뷰트들을 포함하고, 실시예들에 따른 디코더는, 오브젝트 및 상기 뷰포인트들 간 거리에 기반하여 뷰인트들로부터 선택된 특정 개수의 뷰포인트들에 대한 어트리뷰트 데이터를 디코딩할 수 있다.
디코더는, 선택된 특정 개수의 뷰포인트들로부터 생성된 대표 어트리뷰트 정보에 기반하여 특정 개수의 뷰포인트들에 대한 어트리뷰트 데이터를 디코딩할 수 있다.
디코더는, 거리 순서에 기반하여, 선택된 특정 개수의 뷰포인트들에 대한 어트리뷰트 정보로부터 생성된 텍스쳐 데이터를 디코딩할 수 있다.
디코더는, 대표 어트리뷰트 정보 및 특정 개수의 뷰포인트들에 대한 어트리뷰트 정보 간 차이에 기반하여, 특정 개수의 뷰포인트들로부터 생성된 텍스쳐 데이터를 디코딩할 수 있다.
디코더는, 선택된 특정 개수의 뷰포인트들에 대한 어트리뷰트 정보가 머지된 하나의 텍스쳐 데이터를 디코딩할 수 있다.
실시예들에 따른 비트스트림은 카메라 뷰인트 선택에 관련된 파라미터 정보를 포함할 수 있다.
실시예들에 따른 포인트 클라우드 데이터는 오브젝트들로부터 획득된 지오메트리 데이터 및 적어도 두 개의 어트리뷰트들을 포함하고, 실시예들에 따른 비트스트림은 제1오브젝트 및 제2오브젝트 간 가려진 영역에 기반하여 생성된 유효한 카메라 뷰포인트 관련 파라미터 정보를 포함할 수 있다.
이로 인하여, 실시예들에 따른 방법/장치는 정확한 표현을 제공기 위해SLF 데이터 세트를 획득하는 복수 개의 카메라 정보를 모두 압축하고 전송할 필요 없이, 효율적인 데이터 압축 및 복원이 가능하다. 복수 개 중에서 유사도 높은 특정 개수 SLF 데이터만효율적으로 선택하여 압축하기 때문에, 정확하고 압축 성능이 증가한다. 또한, 가려진 영역을 제외하여 데이터를 압축하기 때문에 정확한 데이터 복원이 가능하다.
실시예들은 방법 및/또는 장치 관점에서 설명되었으며, 방법의 설명 및 장치의 설명은 상호 보완하여 적용될 수 있다.
설명의 편의를 위하여 각 도면을 나누어 설명하였으나, 각 도면에 서술되어 있는 실시 예들을 병합하여 새로운 실시 예를 구현하도록 설계하는 것도 가능하다. 그리고, 통상의 기술자의 필요에 따라, 이전에 설명된 실시 예들을 실행하기 위한 프로그램이 기록되어 있는 컴퓨터에서 판독 가능한 기록 매체를 설계하는 것도 실시예들의 권리범위에 속한다. 실시예들에 따른 장치 및 방법은 상술한 바와 같이 설명된 실시 예들의 구성과 방법이 한정되게 적용될 수 있는 것이 아니라, 실시 예들은 다양한 변형이 이루어질 수 있도록 각 실시 예들의 전부 또는 일부가 선택적으로 조합되어 구성될 수도 있다. 실시예들의 바람직한 실시 예에 대하여 도시하고 설명하였지만, 실시예들은 상술한 특정의 실시 예에 한정되지 아니하며, 청구범위에서 청구하는 실시예들의 요지를 벗어남이 없이 당해 발명이 속하는 기술분야에서 통상의 지식을 가진 자에 의해 다양한 변형실시가 가능한 것은 물론이고, 이러한 변형실시들은 실시예들의 기술적 사상이나 전망으로부터 개별적으로 이해돼서는 안 될 것이다.
실시예들의 장치의 다양한 구성요소들은 하드웨어, 소프트웨어, 펌웨어 또는 그것들의 조합에 의해 수행될 수 있다. 실시예들의 다양한 구성요소들은 하나의 칩, 예를 들면 하나의 하드웨어 서킷으로 구현될 수 있다 실시예들에 따라, 실시예들에 따른 구성요소들은 각각 별도의 칩들로 구현될 수 있다. 실시예들에 따라, 실시예들에 따른 장치의 구성요소들 중 적어도 하나 이상은 하나 또는 그 이상의 프로그램들을 실행 할 수 있는 하나 또는 그 이상의 프로세서들로 구성될 수 있으며, 하나 또는 그 이상의 프로그램들은 실시예들에 따른 동작/방법들 중 어느 하나 또는 그 이상의 동작/방법들을 수행시키거나, 수행시키기 위한 인스트럭션들을 포함할 수 있다. 실시예들에 따른 장치의 방법/동작들을 수행하기 위한 실행 가능한 인스트럭션들은 하나 또는 그 이상의 프로세서들에 의해 실행되기 위해 구성된 일시적이지 않은 CRM 또는 다른 컴퓨터 프로그램 제품들에 저장될 수 있거나, 하나 또는 그 이상의 프로세서들에 의해 실행되기 위해 구성된 일시적인 CRM 또는 다른 컴퓨터 프로그램 제품들에 저장될 수 있다. 또한 실시예들에 따른 메모리는 휘발성 메모리(예를 들면 RAM 등)뿐 만 아니라 비휘발성 메모리, 플래쉬 메모리, PROM등을 전부 포함하는 개념으로 사용될 수 있다. 또한, 인터넷을 통한 전송 등과 같은 캐리어 웨이브의 형태로 구현되는 것도 포함될 수 있다. 또한, 프로세서가 읽을 수 있는 기록매체는 네트워크로 연결된 컴퓨터 시스템에 분산되어, 분산방식으로 프로세서가 읽을 수 있는 코드가 저장되고 실행될 수 있다.
이 문서에서 “/”와 “,”는 “및/또는”으로 해석된다. 예를 들어, “A/B”는 “A 및/또는 B”로 해석되고, “A, B”는 “A 및/또는 B”로 해석된다. 추가적으로, “A/B/C”는 “A, B 및/또는 C 중 적어도 하나”를 의미한다. 또한, “A, B, C”도 “A, B 및/또는 C 중 적어도 하나”를 의미한다. 추가적으로, 이 문서에서 “또는”는 “및/또는”으로 해석된다. 예를 들어, “A 또는 B”은, 1) “A” 만을 의미하고, 2) “B” 만을 의미하거나, 3) “A 및 B”를 의미할 수 있다. 달리 표현하면, 본 문서의 “또는”은 “추가적으로 또는 대체적으로(additionally or alternatively)”를 의미할 수 있다.
제1, 제2 등과 같은 용어는 실시예들의 다양한 구성요소들을 설명하기 위해 사용될 수 있다. 하지만 실시예들에 따른 다양한 구성요소들은 위 용어들에 의해 해석이 제한되어서는 안된다. 이러한 용어는 하나의 구성요소를 다른 구성요소와 구별하기 위해 사욛외는 것에 불과하다. 것에 불과하다. 예를 들어, 제1 사용자 인풋 시그널은 제2사용자 인풋 시그널로 지칭될 수 있다. 이와 유사하게, 제2사용자 인풋 시그널은 제1사용자 인풋시그널로 지칭될 수 있다. 이러한 용어의 사용은 다양한 실시예들의 범위 내에서 벗어나지 않는 것으로 해석되어야만 한다. 제1사용자 인풋 시그널 및 제2사용자 인풋 시그널은 모두 사용자 인풋 시그널들이지만, 문맥 상 명확하게 나타내지 않는 한 동일한 사용자 인풋 시그널들을 의미하지 않는다.
실시예들을 설명하기 위해 사용된 용어는 특정 실시예들을 설명하기 위한 목적으로 사용되고, 실시예들을 제한하기 위해서 의도되지 않는다. 실시예들의 설명 및 청구항에서 사용된 바와 같이, 문맥 상 명확하게 지칭하지 않는 한 단수는 복수를 포함하는 것으로 의도된다. 및/또는 표현은 용어 간의 모든 가능한 결합을 포함하는 의미로 사용된다. 포함한다 표현은 특징들, 수들, 단계들, 엘리먼트들, 및/또는 컴포넌트들이 존재하는 것을 설명하고, 추가적인 특징들, 수들, 단계들, 엘리먼트들, 및/또는 컴포넌트들을 포함하지 않는 것을 의미하지 않는다. 실시예들을 설명하기 위해 사용되는, ~인 경우, ~때 등의 조건 표현은 선택적인 경우로만 제한 해석되지 않는다. 특정 조건을 만족하는 때, 특정 조건에 대응하여 관련 동작을 수행하거나, 관련 정의가 해석되도록 의도되었다.
또한, 본 문서에서 설명하는 실시예들에 따른 동작은 실시예들에 따라서 메모리 및/또는 프로세서를 포함하는 송수신 장치에 의해 수행될 수 있다. 메모리는 실시예들에 따른 동작을 처리/제어하기 위한 프로그램들을 저장할 수 있고, 프로세서는 본 문서에서 설명한 다양한 동작을 제어할 수 있다. 프로세서는 컨트롤러 등으로 지칭가능하다. 실시예들에 동작들은 펌웨어, 소프트웨어, 및/또는 그것들의 조합에 의해 수행될 수 있고, 펌웨어, 소프트웨어, 및/또는 그것들의 조합은 프로세서에 저장되거나 메모리에 저장될 수 있다.
상술한 바와 같이, 실시예들을 실시하기 위한 최선의 형태에서 관련 내용을 설명하였다.
상술한 바와 같이, 실시예들은 포인트 클라우드 데이터 송수신 장치 및 시스템에 전체적 또는 부분적으로 적용될 수 있다.
당업자는 실시예들의 범위 내에서 실시예들을 다양하게 변경 또는 변형할 수 있다.
실시예들은 변경/변형들을 포함할 수 있고, 변경/변형은 청구항들 및 그 와 동일한 것들의 범위를 벗어나지 않는다.

Claims (20)

  1. 포인트 클라우드 데이터를 인코딩하는 단계; 및
    상기 포인트 클라우드 데이터를 포함하는 비트스트림을 전송하는 단계; 를 포함하는,
    포인트 클라우드 데이터 송신 방법.
  2. 제1항에 있어서,
    상기 포인트 클라우드 데이터는 뷰포인트들에 대한 카메라들로부터 획득된 지오메트리 데이터 및 적어도 두 개의 어트리뷰트들을 포함하고,
    상기 방법은,
    오브젝트 및 상기 뷰포인트들 간 거리에 기반하여 상기 뷰인트들로부터 특정 개수의 뷰포인트들을 선택하는 단계를 더 포함하는,
    포인트 클라우드 데이터 송신 방법.
  3. 제2항에 있어서, 상기 방법은,
    상기 선택된 특정 개수의 뷰포인트들로부터 대표 어트리뷰트 정보를 생성하고, 상기 대표 어트리뷰트 정보에 기반하여 특정 개수의 뷰포인트들을 선택하는 단계를 더 포함하는,
    포인트 클라우드 데이터 송신 방법.
  4. 제2항에 있어서, 상기 방법은,
    거리 순서에 기반하여, 상기 선택된 특정 개수의 뷰포인트들에 대한 어트리뷰트 정보로부터 텍스쳐 데이터를 생성하는 단계를 더 포함하는,
    포인트 클라우드 데이터 송신 방법.
  5. 제3항에 있어서, 상기 방법은
    상기 대표 어트리뷰트 정보 및 특정 개수의 뷰포인트들에 대한 어트리뷰트 정보 간 차이에 기반하여, 상기 특정 개수의 뷰포인트들로부터 텍스쳐 데이터를 생성하는 단계를 더 포함하는,
    포인트 클라우드 데이터 송신 방법.
  6. 제2항 또는 제3항 중 어느 한 항에 있어서, 상기 방법은,
    상기 선택된 특정 개수의 뷰포인트들에 대한 어트리뷰트 정보를 포함하는 텍스쳐 데이터를 머지하여 하나의 텍스쳐 데이터를 생성하는 단계를 더 포함하는,
    포인트 클라우드 데이터 송신 방법.
  7. 제1항에 있어서,
    상기 비트스트림은 카메라 뷰인트 선택에 관련된 파라미터 정보를 포함하는,
    포인트 클라우드 데이터 송신 방법.
  8. 제1항에 있어서,
    상기 포인트 클라우드 데이터는 오브젝트들로부터 획득된 지오메트리 데이터 및 적어도 두 개의 어트리뷰트들을 포함하고,
    상기 비트스트림은 제1오브젝트 및 제2오브젝트 간 가려진 영역에 기반하여 생성된 유효한 카메라 뷰포인트 관련 파라미터 정보를 포함하는,
    포인트 클라우드 데이터 송신 방법.
  9. 포인트 클라우드 데이터를 포함하는 비트스트림을 수신하는 리시버; 및
    상기 포인트 클라우드 데이터를 디코딩하는 디코더; 를 포함하는,
    포인트 클라우드 데이터 수신 장치.
  10. 제9항에 있어서,
    상기 포인트 클라우드 데이터는 뷰포인트들에 대한 카메라들로부터 획득된 지오메트리 데이터 및 적어도 두 개의 어트리뷰트들을 포함하고,
    상기 디코더는,
    오브젝트 및 상기 뷰포인트들 간 거리에 기반하여 상기 뷰인트들로부터 선택된 특정 개수의 뷰포인트들에 대한 어트리뷰트 데이터를 디코딩하는,
    포인트 클라우드 데이터 수신 장치.
  11. 제10항에 있어서, 상기 디코더는,
    상기 선택된 특정 개수의 뷰포인트들로부터 생성된 대표 어트리뷰트 정보에 기반하여 특정 개수의 뷰포인트들에 대한 어트리뷰트 데이터를 디코딩하는,
    포인트 클라우드 데이터 수신 장치.
  12. 제10항에 있어서, 상기 디코더는,
    거리 순서에 기반하여, 상기 선택된 특정 개수의 뷰포인트들에 대한 어트리뷰트 정보로부터 생성된 텍스쳐 데이터를 디코딩하는,
    포인트 클라우드 데이터 수신 장치.
  13. 제11항에 있어서, 상기 디코더는,
    상기 대표 어트리뷰트 정보 및 특정 개수의 뷰포인트들에 대한 어트리뷰트 정보 간 차이에 기반하여, 상기 특정 개수의 뷰포인트들로부터 생성된 텍스쳐 데이터를 디코딩하는,
    포인트 클라우드 데이터 수신 장치.
  14. 제10항 또는 제11항 중 어느 한 항에 있어서, 상기 디코더는,
    상기 선택된 특정 개수의 뷰포인트들에 대한 어트리뷰트 정보가 머지된 하나의 텍스쳐 데이터를 디코딩하는,
    포인트 클라우드 데이터 수신 장치.
  15. 제9항에 있어서,
    상기 비트스트림은 카메라 뷰인트 선택에 관련된 파라미터 정보를 포함하는,
    포인트 클라우드 데이터 수신 장치.
  16. 제9항에 있어서,
    상기 포인트 클라우드 데이터는 오브젝트들로부터 획득된 지오메트리 데이터 및 적어도 두 개의 어트리뷰트들을 포함하고,
    상기 비트스트림은 제1오브젝트 및 제2오브젝트 간 가려진 영역에 기반하여 생성된 유효한 카메라 뷰포인트 관련 파라미터 정보를 포함하는,
    포인트 클라우드 데이터 수신 장치.
  17. 포인트 클라우드 데이터를 인코딩하는 인코더; 및
    상기 포인트 클라우드 데이터를 포함하는 비트스트림을 전송하는 트랜스미터; 를 포함하는,
    포인트 클라우드 데이터 송신 장치.
  18. 제17항에 있어서,
    상기 포인트 클라우드 데이터는 뷰포인트들에 대한 카메라들로부터 획득된 지오메트리 데이터 및 적어도 두 개의 어트리뷰트들을 포함하고,
    상기 인코더는
    오브젝트 및 상기 뷰포인트들 간 거리에 기반하여 상기 뷰인트들로부터 특정 개수의 뷰포인트들을 선택하는,
    포인트 클라우드 데이터 송신 장치.
  19. 포인트 클라우드 데이터를 포함하는 비트스트림을 수신하는 단계; 및
    상기 포인트 클라우드 데이터를 디코딩하는 단계; 를 포함하는,
    포인트 클라우드 데이터 수신 방법.
  20. 제19항에 있어서,
    상기 포인트 클라우드 데이터는 뷰포인트들에 대한 카메라들로부터 획득된 지오메트리 데이터 및 적어도 두 개의 어트리뷰트들을 포함하고,
    상기 디코더하는 단계는,
    오브젝트 및 상기 뷰포인트들 간 거리에 기반하여 상기 뷰인트들로부터 선택된 특정 개수의 뷰포인트들에 대한 어트리뷰트 데이터를 디코딩하는,
    포인트 클라우드 데이터 수신 방법.
PCT/KR2021/011600 2020-09-11 2021-08-30 포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법 Ceased WO2022055165A1 (ko)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US18/022,900 US20230419557A1 (en) 2020-09-11 2021-08-30 Point cloud data transmission device, point cloud data transmission method, point cloud data reception device, and point cloud data reception method

Applications Claiming Priority (4)

Application Number Priority Date Filing Date Title
KR10-2020-0117182 2020-09-11
KR20200117182 2020-09-11
KR20200127235 2020-09-29
KR10-2020-0127235 2020-09-29

Publications (1)

Publication Number Publication Date
WO2022055165A1 true WO2022055165A1 (ko) 2022-03-17

Family

ID=80632264

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/KR2021/011600 Ceased WO2022055165A1 (ko) 2020-09-11 2021-08-30 포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법

Country Status (2)

Country Link
US (1) US20230419557A1 (ko)
WO (1) WO2022055165A1 (ko)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2024149396A1 (en) * 2023-01-14 2024-07-18 Douyin Vision Co., Ltd. Method, apparatus, and medium for video processing
WO2024149182A1 (zh) * 2023-01-11 2024-07-18 维沃移动通信有限公司 变换系数编码方法、变换系数解码方法及终端

Families Citing this family (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP7809720B2 (ja) * 2021-04-07 2026-02-02 インターディジタル・シーイー・パテント・ホールディングス・ソシエテ・パ・アクシオンス・シンプリフィエ 光効果をサポートする容積ビデオ
US11900525B2 (en) * 2022-02-14 2024-02-13 Google Llc Learned volumetric attribute compression using coordinate-based networks
US12363343B2 (en) * 2022-07-14 2025-07-15 Apple Inc. Base mesh data and motion information sub-stream format for video-based dynamic mesh compression
US20240291956A1 (en) * 2023-02-28 2024-08-29 Nokia Technologies Oy Method, an apparatus and a computer program product for real-time streaming of volumetric video
US12513341B2 (en) * 2023-05-11 2025-12-30 Sharp Kabushiki Kaisha 3D video decoding apparatus and 3D video coding apparatus
US12573170B2 (en) * 2023-09-19 2026-03-10 International Business Machines Corporation Unsupervised video segment clustering and boundary positioning
US12548249B2 (en) * 2023-12-12 2026-02-10 Qualcomm Incorporated Incremental surface completion
US20250234085A1 (en) * 2024-01-16 2025-07-17 Samsung Electronics Co., Ltd. Sei message and track for 2d snapshot image of compessed volumetric contents for quick preview or trick play operations

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2019135024A1 (en) * 2018-01-02 2019-07-11 Nokia Technologies Oy An apparatus, a method and a computer program for volumetric video
KR20200005438A (ko) * 2018-07-06 2020-01-15 엘지전자 주식회사 동적 뷰포인트의 좌표계에 대한 메타데이터를 송수신하는 방법 및 장치
KR20200038170A (ko) * 2018-10-02 2020-04-10 엘지전자 주식회사 360 비디오 시스템에서 오버레이 처리 방법 및 그 장치
KR20200038534A (ko) * 2017-09-18 2020-04-13 애플 인크. 포인트 클라우드 압축
KR20200065076A (ko) * 2017-10-20 2020-06-08 인터디지털 브이씨 홀딩스 인코포레이티드 볼류메트릭 비디오 포맷을 위한 방법, 장치 및 스트림

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20200153885A1 (en) * 2018-10-01 2020-05-14 Lg Electronics Inc. Apparatus for transmitting point cloud data, a method for transmitting point cloud data, an apparatus for receiving point cloud data and/or a method for receiving point cloud data
US11831861B2 (en) * 2019-08-12 2023-11-28 Intel Corporation Methods for viewport-dependent adaptive streaming of point cloud content

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20200038534A (ko) * 2017-09-18 2020-04-13 애플 인크. 포인트 클라우드 압축
KR20200065076A (ko) * 2017-10-20 2020-06-08 인터디지털 브이씨 홀딩스 인코포레이티드 볼류메트릭 비디오 포맷을 위한 방법, 장치 및 스트림
WO2019135024A1 (en) * 2018-01-02 2019-07-11 Nokia Technologies Oy An apparatus, a method and a computer program for volumetric video
KR20200005438A (ko) * 2018-07-06 2020-01-15 엘지전자 주식회사 동적 뷰포인트의 좌표계에 대한 메타데이터를 송수신하는 방법 및 장치
KR20200038170A (ko) * 2018-10-02 2020-04-10 엘지전자 주식회사 360 비디오 시스템에서 오버레이 처리 방법 및 그 장치

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2024149182A1 (zh) * 2023-01-11 2024-07-18 维沃移动通信有限公司 变换系数编码方法、变换系数解码方法及终端
WO2024149396A1 (en) * 2023-01-14 2024-07-18 Douyin Vision Co., Ltd. Method, apparatus, and medium for video processing

Also Published As

Publication number Publication date
US20230419557A1 (en) 2023-12-28

Similar Documents

Publication Publication Date Title
WO2021066615A1 (en) Point cloud data transmission device, point cloud data transmission method, point cloud data reception device, and point cloud data reception method
WO2020190075A1 (ko) 포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법
WO2021002633A2 (ko) 포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법
WO2021187737A1 (ko) 포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법
WO2021002657A1 (ko) 포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법
WO2021141264A1 (ko) 포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법
WO2020190114A1 (ko) 포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법
WO2022055165A1 (ko) 포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법
WO2021071257A1 (ko) 포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법
WO2021141258A1 (ko) 포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법
WO2020189895A1 (ko) 포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법
WO2021261865A1 (ko) 포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법
WO2021141233A1 (ko) 포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법
WO2021141208A1 (ko) 포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법
WO2021210763A1 (ko) 포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법
WO2021210860A1 (ko) 포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법
WO2021002730A1 (ko) 포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법
WO2020189903A1 (ko) 포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법
WO2021025251A1 (ko) 포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법
WO2021206333A1 (ko) 포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법
WO2021261897A1 (ko) 포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법
WO2021256909A1 (ko) 포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법
WO2023167430A1 (ko) 포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법
WO2021002558A1 (ko) 포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법.
WO2021206365A1 (ko) 포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21867031

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21867031

Country of ref document: EP

Kind code of ref document: A1