CN115474047A - LiDAR point cloud encoding method and decoding method based on enhanced map correlation - Google Patents
LiDAR point cloud encoding method and decoding method based on enhanced map correlation Download PDFInfo
- Publication number
- CN115474047A CN115474047A CN202211112366.9A CN202211112366A CN115474047A CN 115474047 A CN115474047 A CN 115474047A CN 202211112366 A CN202211112366 A CN 202211112366A CN 115474047 A CN115474047 A CN 115474047A
- Authority
- CN
- China
- Prior art keywords
- frame
- point cloud
- map
- frames
- distance map
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 39
- 238000009616 inductively coupled plasma Methods 0.000 claims abstract description 38
- 238000013507 mapping Methods 0.000 claims abstract description 18
- 239000011159 matrix material Substances 0.000 claims description 36
- 238000013519 translation Methods 0.000 claims description 18
- 230000009466 transformation Effects 0.000 claims description 9
- 238000013139 quantization Methods 0.000 claims description 7
- 238000007906 compression Methods 0.000 abstract description 5
- 230000006835 compression Effects 0.000 abstract description 5
- 238000011161 development Methods 0.000 abstract description 3
- 238000011160 research Methods 0.000 abstract description 2
- 238000012545 processing Methods 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000000903 blocking effect Effects 0.000 description 1
- 238000013144 data compression Methods 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/169—Methods 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
- H04N19/17—Methods 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 an image region, e.g. an object
- H04N19/172—Methods 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 an image region, e.g. an object the region being a picture, frame or field
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
- G06T7/32—Determination of transform parameters for the alignment of images, i.e. image registration using correlation-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T9/00—Image coding
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods 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
- H04N19/124—Quantisation
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
- H04N19/597—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding specially adapted for multi-view video sequence encoding
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/70—Methods 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
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/90—Methods 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
- H04N19/91—Entropy coding, e.g. variable length coding [VLC] or arithmetic coding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10028—Range image; Depth image; 3D point clouds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20021—Dividing image into blocks, subimages or windows
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The invention relates to a LiDAR point cloud encoding method and a LiDAR point cloud decoding method based on enhanced mapping graph correlation, wherein the encoding method comprises the following steps: acquiring point clouds to be encoded, taking part of point cloud frames as I frames, and taking the rest of point cloud frames as P frames; mapping the geometric information of I and P frames into a 2D distance map, divided into N resolution 1 ×N 2 Is divided into blocks, I frames andconverting each block of the P frame into three-dimensional point cloud, and registering each block of the P frame with the three-dimensional point cloud of each block of the I frame by using an ICP (inductively coupled plasma) algorithm to obtain motion information H; converting each blocked three-dimensional point cloud of the P' frame into a 2D distance map; mapping the attribute information of the I frame and the P frame into a 2D attribute graph; encoding the 2D distance graph and the 2D attribute graph into a video stream, and encoding the video stream into a binary code stream; the motion information H is quantized into a 2D distance map and compressed into a binary code stream. The invention reduces the space and time redundancy, and has important strategic significance for the exploration and research of a high-efficiency point cloud compression scheme on the development of relevant applications such as automatic driving, mobile robots and the like.
Description
Technical Field
The invention relates to a LiDAR point cloud encoding method and a LiDAR point cloud decoding method based on enhanced map correlation, and belongs to the technical field of point cloud processing.
Background
The development of three-dimensional sensor data acquisition generates new opportunities and challenges in the industries such as virtual reality technology, smart cities, intelligent robots and automatic driving, an automatic driving system has wide application in the fields of reducing road traffic working pressure, improving natural environments of traffic trips and the like, and the identification of the surrounding environment and the state of a vehicle by using a laser radar is one of the bases of automatic driving.
Because the three-dimensional point cloud acquisition has the advantages of high efficiency, high precision, high resolution, digital acquisition and the like, a plurality of automatic driving systems including Google and Uber are practically developed and applied on the basis of 3D point cloud data acquired by a vehicle-mounted laser radar. The point cloud model represents a visual scene by a series of three-dimensional points, describing topological information of the object surface and the associated attributes of the points. However, the non-uniform sampling and spatial structural distribution disorder of the point cloud make the point cloud compression and processing quite challenging, for example, the 64-line radar sensor produced by Velodyne has more than one hundred thousand points collected by one rotation, and under normal setting, the data of the sensor in half an hour is as high as more than 50G. Therefore, the amount of three-dimensional data generated by the point cloud acquisition device presents a higher challenge to the storage and transmission of data. In recent years, the three-dimensional point cloud data compression technology is gradually and deeply researched, and the inventor finds that the three-dimensional point cloud has the problems of wide point cloud distribution range and inconsistent density.
Disclosure of Invention
In order to overcome the problems, the invention provides a LiDAR point cloud encoding method and a LiDAR point cloud decoding method based on the relevance of an enhanced mapping map, the encoding method reduces the spatial and temporal redundancy, and has important strategic significance for the exploration and research of an efficient point cloud compression scheme on the development of relevant applications such as automatic driving and mobile robots.
The detachable electromechanical mounting frame is simple in structure, convenient to detach, replaceable in parts and suitable for mounting electromechanical equipment of different specifications.
The technical scheme of the invention is as follows:
first aspect of the invention
A LiDAR point cloud coding method based on enhanced map correlation comprises the following steps:
acquiring a point cloud to be encoded, taking a part of point cloud frames as I frames, and taking the rest point cloud frames as P frames;
mapping the geometric information of the I frame and the P frame into a 2D distance map, and dividing the 2D distance map into N resolution ratios 1 ×N 2 Partition of (2), N 1 Converting the blocks of the I frame and the P frame into three-dimensional point clouds for the line number of the laser radar, and registering the block three-dimensional point clouds of the P frame according to the block three-dimensional point clouds of the I frame by using an ICP (inductively coupled plasma) algorithm to obtain the registered motion information H of the block three-dimensional point clouds of the P' frame and the block three-dimensional point clouds of the P frame on the block three-dimensional point clouds of the I frame; the motion information H comprises a rotation matrix and a translation matrix;
Converting the partitioned three-dimensional point clouds of the P' frame into a 2D distance map;
mapping the attribute information of the I frame and the P frame into a 2D attribute graph;
respectively encoding the 2D distance map of the I frame, the 2D distance map of the P' frame and the 2D attribute map into a video stream, and encoding the video stream into a binary code stream through HEVC;
the motion information H is quantized into a 2D distance map and compressed into a binary code stream by entropy coding.
Further, N 2 And N 1 The same is true.
Further, the geometric information is mapped to a 2D distance map, specifically:
from three-dimensional pointsYaw angle omega of cloud data computing point i i Angle of pitch theta i And distance r of point i from the center i The disclosure is as follows:
wherein x is i Is the left value of the x axis of the point i Cartesian coordinate, y i Is the left value of the point i Cartesian coordinate y axis; z is a radical of i The left value of the Z axis of the Cartesian coordinate of the point i is obtained;
will point (r) i ，θ i ，ω i ) Mapping to pixel P (px) i ，py i ) The formula is as follows:
P(px i ,py i )＝round(r i )；
wherein x _ quan and y _ quan are the yaw angle ω respectively i And the pitch angle theta i Quantized into integers;
according to the pixel P (px) i ，py i ) A 2D distance map is obtained.
Further, the set information is mapped to a 2D distance map of 16-bit depth values.
Further, the ICP algorithm is an ICP improvement algorithm, and the ICP improvement algorithm specifically comprises:
Estimating motion information between point cloud frames according to an ICP improved algorithm based on a translation model, wherein the formula is as follows:
wherein f (R, t) is an error function value of unpermitted punctual point clouds, fmin (R, t) is an error function of motion information after registration, n is the total number of source point set point clouds, and p i Is an ICP source point set, t is a translation matrix, q i R is a rotation matrix for the ICP target point set.
Further, the ICP improvement algorithm further comprises:
after the ICP improvement algorithm iterates N times, let the rotation matrix equal the identity matrix, i.e. R = R0:
and solving the translation matrix, and using an ICP algorithm for the newly obtained translation matrix and the unit matrix.
Further, the I-frame 2D distance map, the P' frame 2D distance map, and the 2D attribute map are respectively encoded into a video stream, and the video stream is encoded into a binary code stream by HEVC, specifically:
and coding the I frame 2D distance map, the P' frame 2D distance map and the 2D attribute map into a video stream, and compressing the video stream by using a video coding method to obtain a binary code stream.
Further, quantizing the motion information H into a 2D distance map specifically includes:
the motion information H is quantized in a uniform quantization manner, and the formula is as follows:
wherein, P m (i, j) is the quantized pixel, H (i, j) is the rigid transform matrix H in i rows and j columnsNumber H m in (i, j) is the minimum value in the rigid transformation matrix H, H max (i, j) is the maximum value in the rigid transformation matrix H.
Further, the compressing into a binary code stream by entropy coding specifically includes:
performing 4-level 5/3 wavelet transform on the 2D distance map, and calculating the optimal quantization step size of each sub-band;
removing related redundancy among pixels of the data of the lowest-frequency sub-band by adopting a JPEG-LS algorithm, and coding by using zero run coding to obtain a zero run coding result;
and carrying out context modeling on the zero-run-length encoding result of the pattern, updating the number of levels of direct-tree-Columbus encoding, and obtaining a binary code stream by using the direct-tree-Columbus encoding of the corresponding number of levels.
Second aspect of the invention
A decoding method based on enhanced mapping map correlation LiDAR point cloud coding is used for decoding and recovering point clouds coded by the enhanced mapping map correlation LiDAR point cloud coding method in the first aspect into original point clouds, and comprises the following steps:
decoding motion information, namely performing entropy decoding on the binary code stream obtained by entropy coding to obtain a 2D distance map of the P 'frame, and converting registration information in the 2D distance map of the P' frame into motion information H;
Decoding geometric information, namely performing HEVC decoding on a binary code stream obtained by HEVC coding to obtain a 2D distance map of an I frame and a 2D distance map of a P 'frame containing the geometric information, converting the 2D distance maps of the I frame and the P' frame into three-dimensional point clouds of the I frame and the P 'frame, and restoring the three-dimensional point clouds of the P' frame by using decoded motion information H to obtain the geometric information of the P frame;
decoding attribute information, namely performing HEVC decoding on a binary code stream obtained by HEVC coding to obtain a 2D attribute graph containing the attribute information, and converting the 2D attribute graphs of I frames and P frames into the attribute information of the I frames and P' frames;
and recovering the three-dimensional point clouds of the I frame and the P frame according to the 2D distance map, the 2D attribute map, the geometric information and the attribute information.
The invention has the following beneficial effects:
adopting a method of mapping point cloud data to a distance map and an attribute map; in order to improve the time correlation of the distance map of the point cloud, a point cloud blocking and registration method is provided; in order to improve the accuracy of registration, a translation model is provided to improve the registration method of ICP; and converting the point cloud after registration into a distance map again, forming a video sequence by using the distance map and the attribute graph, coding by using HEVC, and mapping the motion information required by registration into a motion map by using entropy coding.
Drawings
FIG. 1 is a flow chart of the encoding method of the present invention.
Fig. 2 is a graph of the ICP improvement algorithm of the present invention.
FIG. 3 is a flow chart of the adaptive exponential Golomb coding algorithm of the present invention.
FIG. 4 is a flowchart of a decoding method according to the present invention.
Detailed Description
The invention is described in detail below with reference to the figures and the specific embodiments.
First aspect
Example one
Referring to FIG. 1, a LiDAR point cloud encoding method based on enhanced map correlation includes the following steps:
acquiring a point cloud to be encoded, taking a part of point cloud frames as I frames, and taking the rest point cloud frames as P frames;
mapping the geometrical information of the I frame and the P frame into a 2D distance map, and dividing the 2D distance map into N frames with the resolution 1 ×N 2 Is divided into blocks of N 1 Converting the blocks of the I frame and the P frame into three-dimensional point clouds for the line number of the laser radar, and registering the block three-dimensional point clouds of the P frame according to the block three-dimensional point clouds of the I frame by using an ICP (inductively coupled plasma) algorithm to obtain the registered motion information H of the block three-dimensional point clouds of the P' frame and the block three-dimensional point clouds of the P frame on the block three-dimensional point clouds of the I frame; the motion information comprises a rotation matrix and a translation matrix;
converting the partitioned three-dimensional point clouds of the P' frame into a 2D distance map;
Mapping the attribute information of the I frame and the P frame into a 2D attribute graph;
respectively encoding the 2D distance map of the I frame, the 2D distance map of the P' frame and the 2D attribute map into video streams, and encoding the video streams into binary code streams through HEVC;
and quantizing the motion information H into a 2D distance map, and compressing the distance map into a binary code stream through entropy coding.
The invention appoints a part of point cloud frame as reference frame, also called I frame (intra-frame coding frame), belonging to key frame; and uses them to predict other non-key point cloud frames, called P-frames (inter-frame predicted frames).
In one embodiment of the invention, in order to make the point number of point cloud registration at a reasonable and stable value, consider N from the experimental experience perspective 2 And N 1 The same is true.
Example two
On the basis of the first embodiment, the geometric information is mapped into a 2D distance map, specifically:
calculating the yaw angle omega of the point i according to the three-dimensional point cloud data i Angle of pitch theta i And distance r of point i from the center i The publication is as follows:
wherein x is i Is the left value of the x axis of the point i Cartesian coordinate, y i Is the left value of the point i Cartesian coordinate y axis; z is a radical of formula i The left value of the Z axis of the Cartesian coordinate of the point i is obtained;
will point (r) i ，θ i ，ω i ) Mapping to pixel P (px) i ，py i ) The formula is as follows:
P(px i ,py i )＝round(r i )；
wherein x _ quan and y _ quan are the yaw angle ω respectively i And the pitch angle theta i Quantized into integers;
according to the pixel P (px) i ，py i ) A 2D distance map is obtained.
The pitch angle is determined by the inherent resolution of a vehicle-mounted laser radar, for example, a 16-line, 32-line and 64-line laser radar, the yaw angle is generally more than one point from 0 to 360 degrees, the effective sampling distance of the laser radar is generally no more than 130 meters at present, because the original data is obtained by measuring the radial distance from the center of the laser radar to an object, an index taking the pitch angle as a row and an index taking the yaw angle as a column are obtained, the distance value corresponding to a certain pitch angle and the yaw angle is the pixel value of a gray scale image, and the original compressed disordered scattered point cloud compression is converted into the compression of the distance image (gray scale image) with stronger correlation.
In one embodiment of the invention, the set information is mapped as a 2D distance map of 16-bit depth values.
EXAMPLE III
On the basis of the first embodiment, the ICP algorithm is an ICP improvement algorithm, and the ICP improvement algorithm specifically comprises the following steps:
Estimating motion information between point cloud frames according to an ICP improved algorithm based on a translation model, wherein the formula is as follows:
wherein f (R, t) is an error function value of unpermitted punctual point clouds, fmin (R, t) is an error function of motion information after registration, n is the total number of source point set point clouds, and p i Is an ICP source point set, t is a translation matrix, q i And R is a rotation matrix for the ICP target point set.
The ICP algorithm is a rigid transformation algorithm using a wide application in point cloud registration, and the present invention estimates motion information from frame to frame of a point cloud by using a translation model-based ICP improvement algorithm.
The point cloud is partitioned and the corresponding modules are registered, so that the registration accuracy can be further improved, the operation speed can be reduced, as the laser radar point cloud is sparse and the acquisition unit rotates for one circle vertical to the vehicle, a large number of macro blocks have no data by using octree partitioning, and a large amount of encoded data and time are wasted by encoding and indexing the macro blocks with point cloud geometric data of each frame.
Referring to fig. 2, in one embodiment of the present invention, the ICP improvement algorithm further comprises:
after the ICP improvement algorithm iterates N times, let the rotation matrix equal the identity matrix, i.e., R = R0:
and solving the translation matrix, and using an ICP algorithm for the newly obtained translation matrix and the unit matrix.
According to the collected data characteristics of the point cloud of the vehicle-mounted laser radar: the acquisition time interval between frames is short, and the road surface is smooth. At this time, the operation information between frames is mainly a large translation vector and a small rotation vector. Therefore, the method not only ensures the approach to the true value, but also can improve the coding efficiency.
Example four
On the basis of the first embodiment, the I-frame 2D distance map, the P' frame 2D distance map, and the 2D attribute map are respectively encoded into a video stream, and the video stream is encoded into a binary code stream by HEVC, specifically:
and coding the I frame 2D distance map, the P' frame 2D distance map and the 2D attribute map into a video stream, and compressing the video stream by using a video coding method to obtain a binary code stream.
EXAMPLE five
A LiDAR point cloud coding method based on enhanced map correlation is disclosed, and on the basis of the first embodiment, the quantization of the motion information H into a 2D distance map specifically comprises the following steps:
the motion information H is quantized in a uniform quantization manner, and the formula is as follows:
wherein, P m (i, j) is the quantized pixel, H (i, j) is the value of the rigid transformation matrix H in i rows and j columns, H min (i, j) is the minimum value in the rigid transformation matrix H, H max (i, j) is the maximum value in the rigid transformation matrix H.
EXAMPLE six
Referring to fig. 3, a LiDAR point cloud encoding method based on enhanced map correlation is, on the basis of the first embodiment, compressed into a binary code stream by entropy encoding, specifically:
performing 4-level 5/3 wavelet transform on the 2D distance map, and calculating the optimal quantization step size of each sub-band;
removing related redundancy among pixels of the data of the lowest-frequency sub-band by adopting a JPEG-LS algorithm, and coding by using zero run coding to obtain a zero run coding result;
and carrying out context modeling on the zero-run-length encoding result of the pattern, updating the number of levels of direct-tree-Columbus encoding, and obtaining a binary code stream by using the direct-tree-Columbus encoding of the corresponding number of levels.
Second aspect of the invention
EXAMPLE seven
Referring to fig. 4, a method for decoding LiDAR point cloud encoding based on enhanced map correlation, which decodes the point cloud encoded by the LiDAR point cloud encoding based on enhanced map correlation according to the first aspect to recover the original point cloud, and includes the following steps:
decoding motion information, namely performing entropy decoding on the binary code stream obtained by entropy coding to obtain a 2D distance graph of the P 'frame, and converting registration information in the 2D distance graph of the P' frame into motion information H;
Decoding geometric information, namely performing HEVC decoding on binary code stream obtained by HEVC coding to obtain a 2D distance map of an I frame and a 2D distance map of a P 'frame containing the geometric information, converting the 2D distance maps of the I frame and the P' frame into three-dimensional point clouds of the I frame and the P 'frame, and restoring the three-dimensional point clouds of the P' frame by using decoded motion information H to obtain the geometric information of the P frame;
decoding attribute information, namely performing HEVC decoding on a binary code stream obtained by HEVC coding to obtain a 2D attribute graph containing the attribute information, and converting the 2D attribute graphs of I frames and P frames into the attribute information of the I frames and P' frames;
and recovering the three-dimensional point clouds of the I frame and the P frame according to the 2D distance map, the 2D attribute map, the geometric information and the attribute information.
The above description is only an embodiment of the present invention, and not intended to limit the scope of the present invention, and all equivalent structures made by using the contents of the specification and the drawings of the present invention or directly or indirectly applied to other related technical fields are included in the scope of the present invention.
Claims (10)
1. A LiDAR point cloud coding method based on enhanced map correlation is characterized by comprising the following steps:
acquiring a point cloud to be encoded, taking a part of point cloud frames as I frames, and taking the rest point cloud frames as P frames;
Mapping the geometric information of the I frame and the P frame into a 2D distance map, and dividing the 2D distance map into N resolution ratios 1 ×N 2 Is divided into blocks of N 1 Converting the blocks of the I frame and the P frame into three-dimensional point clouds for the line number of the laser radar, and registering the block three-dimensional point clouds of the P frame according to the block three-dimensional point clouds of the I frame by using an ICP (inductively coupled plasma) algorithm to obtainMotion information H of each block point cloud of the registered P' frame and each block three-dimensional point cloud of the P frame to each block three-dimensional point cloud of the I frame; the motion information H comprises a rotation matrix and a translation matrix;
converting the partitioned three-dimensional point clouds of the P' frame into a 2D distance map;
mapping the attribute information of the I frame and the P frame into a 2D attribute graph;
respectively encoding the 2D distance map of the I frame, the 2D distance map of the P' frame and the 2D attribute map into a video stream, and encoding the video stream into a binary code stream through HEVC;
the motion information H is quantized into a 2D distance map and compressed into a binary code stream by entropy coding.
2. The LiDAR point cloud encoding method based on enhanced map correlation of claim 1, wherein N is 2 And N 1 The same is true.
3. The LiDAR point cloud encoding method based on enhanced map correlation according to claim 1, wherein the geometric information is mapped as a 2D distance map, specifically:
Calculating the yaw angle omega of the point i according to the three-dimensional point cloud data i Pitch angle θ i Distance r of point i from the center i The publication is as follows:
wherein x is i Is the left value of the x axis of the point i Cartesian coordinate, y i Is the left value of the point i Cartesian coordinate y axis; z is a radical of formula i Is the value of the left side of the z axis of the point i Cartesian coordinate;
will point (r) i ，θ i ，ω i ) Mapping to pixel P (px) i ，py i ) The formula is as follows:
P(px i ，py i )＝round(r i )；
wherein x _ quan and y _ quan are the yaw angle ω respectively i And the pitch angle theta i Quantized into integers;
according to the pixel P (px) i ，py i ) A 2D distance map is obtained.
4. The enhanced map correlation-based LiDAR point cloud encoding method of claim 3, wherein the set information is mapped as a 2D distance map of 16-bit depth values.
5. The LiDAR point cloud encoding method based on enhanced map correlation according to claim 1, wherein the ICP algorithm is an ICP improvement algorithm, and the ICP improvement algorithm is specifically:
estimating motion information between point cloud frames according to an ICP improved algorithm based on a translation model, wherein the formula is as follows:
wherein f (R, t) is an error function value of unpermitted punctual point clouds, fmin (R, t) is an error function of motion information after registration, n is the total number of source point set point clouds, and p i Is an ICP source point set, t is a translation matrix, q i And R is a rotation matrix for the ICP target point set.
6. The enhanced map correlation-based LiDAR point cloud encoding method of claim 6, wherein the ICP-refinement algorithm further comprises:
after the ICP improvement algorithm iterates N times, let the rotation matrix equal the identity matrix, i.e., R = R0:
and solving the translation matrix, and using an ICP algorithm for the newly obtained translation matrix and the unit matrix.
7. The enhanced map correlation-based LiDAR point cloud encoding method of claim 1, wherein the I-frame 2D distance map, the P' frame 2D distance map, and the 2D attribute map are encoded into video streams, respectively, and the video streams are encoded into binary code streams by HEVC, specifically:
and coding the I frame 2D distance map, the P' frame 2D distance map and the 2D attribute map into a video stream, and compressing the video stream by using a video coding method to obtain a binary code stream.
8. The LiDAR point cloud encoding method based on enhanced map correlation according to claim 1, wherein quantizing the motion information H into a 2D distance map specifically is:
and quantizing the motion information H in a uniform quantization mode, wherein the formula is as follows:
Wherein, P m (i, j) is the quantized pixel, H (i, j) is the value of the rigid transformation matrix H in i rows and j columns, H min (i, j) in rigid transformation matrix HMinimum value, H max (i, j) is the maximum value in the rigid transformation matrix H.
9. The enhanced map correlation-based LiDAR point cloud encoding method of claim 7, wherein the encoding is compressed into a binary code stream by entropy encoding, specifically:
performing 4-level 5/3 wavelet transform on the 2D distance map, and calculating the optimal quantization step size of each sub-band;
removing related redundancy among pixels of the data of the lowest-frequency sub-band by adopting a JPEG-LS algorithm, and coding by using zero run coding to obtain a zero run coding result;
and carrying out context modeling on the zero-run-length encoding result of the pattern, updating the number of levels of direct-tree-Columbus encoding, and obtaining a binary code stream by using the direct-tree-Columbus encoding of the corresponding number of levels.
10. A decoding method based on enhanced map correlation LiDAR point cloud encoding, which decodes the point cloud encoded by the enhanced map correlation LiDAR point cloud encoding method of claims 1-8 to recover the original point cloud, comprising the following steps:
decoding motion information, namely performing entropy decoding on the binary code stream obtained by entropy coding to obtain a 2D distance graph of the P 'frame, and converting registration information in the 2D distance graph of the P' frame into motion information H;
Decoding geometric information, namely performing HEVC decoding on a binary code stream obtained by HEVC coding to obtain a 2D distance map of an I frame and a 2D distance map of a P 'frame containing the geometric information, converting the 2D distance maps of the I frame and the P' frame into three-dimensional point clouds of the I frame and the P 'frame, and restoring the three-dimensional point clouds of the P' frame by using decoded motion information H to obtain the geometric information of the P frame;
decoding attribute information, namely performing HEVC decoding on binary code streams obtained by HEVC coding to obtain a 2D attribute graph containing attribute information, and converting the 2D attribute graphs of I frames and P frames into attribute information of the I frames and the P' frames;
and recovering the three-dimensional point clouds of the I frame and the P frame according to the 2D distance map, the 2D attribute map, the geometric information and the attribute information.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211112366.9A CN115474047A (en) | 2022-09-13 | 2022-09-13 | LiDAR point cloud encoding method and decoding method based on enhanced map correlation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211112366.9A CN115474047A (en) | 2022-09-13 | 2022-09-13 | LiDAR point cloud encoding method and decoding method based on enhanced map correlation |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115474047A true CN115474047A (en) | 2022-12-13 |
Family
ID=84332785
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211112366.9A Pending CN115474047A (en) | 2022-09-13 | 2022-09-13 | LiDAR point cloud encoding method and decoding method based on enhanced map correlation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115474047A (en) |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110688905A (en) * | 2019-08-30 | 2020-01-14 | 中山大学 | Three-dimensional object detection and tracking method based on key frame |
| CN110832856A (en) * | 2017-11-30 | 2020-02-21 | 深圳市大疆创新科技有限公司 | System and method for reducing video coding fluctuations |
| CN112334949A (en) * | 2018-10-09 | 2021-02-05 | 松下电器（美国）知识产权公司 | Three-dimensional data encoding method, three-dimensional data decoding method, three-dimensional data encoding device, and three-dimensional data decoding device |
| CN113382242A (en) * | 2021-05-20 | 2021-09-10 | 北京大学深圳研究生院 | Code rate control method, device, equipment and computer readable storage medium |
| CN113557741A (en) * | 2019-03-08 | 2021-10-26 | Vid拓展公司 | Method and apparatus for adaptive streaming of point clouds |
| CN113573068A (en) * | 2021-07-28 | 2021-10-29 | 福州大学 | Improved V-PCC (V-PCC) inter-frame prediction method and system based on registration |
| US20210407149A1 (en) * | 2019-03-25 | 2021-12-30 | Panasonic Intellectual Property Corporation Of America | Three-dimensional data encoding method, three-dimensional data decoding method, three-dimensional data encoding device, and three-dimensional data decoding device |
| WO2022042538A1 (en) * | 2020-08-24 | 2022-03-03 | 北京大学深圳研究生院 | Block-based point cloud geometric inter-frame prediction method and decoding method |
| CN114332259A (en) * | 2021-12-29 | 2022-04-12 | 福州大学 | Point cloud coding and decoding method based on vehicle-mounted laser radar |
| CN114663600A (en) * | 2022-04-18 | 2022-06-24 | 南京大学 | Point cloud reconstruction method and system based on self-encoder |
-
2022
- 2022-09-13 CN CN202211112366.9A patent/CN115474047A/en active Pending
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110832856A (en) * | 2017-11-30 | 2020-02-21 | 深圳市大疆创新科技有限公司 | System and method for reducing video coding fluctuations |
| CN112334949A (en) * | 2018-10-09 | 2021-02-05 | 松下电器（美国）知识产权公司 | Three-dimensional data encoding method, three-dimensional data decoding method, three-dimensional data encoding device, and three-dimensional data decoding device |
| CN113557741A (en) * | 2019-03-08 | 2021-10-26 | Vid拓展公司 | Method and apparatus for adaptive streaming of point clouds |
| US20210407149A1 (en) * | 2019-03-25 | 2021-12-30 | Panasonic Intellectual Property Corporation Of America | Three-dimensional data encoding method, three-dimensional data decoding method, three-dimensional data encoding device, and three-dimensional data decoding device |
| CN110688905A (en) * | 2019-08-30 | 2020-01-14 | 中山大学 | Three-dimensional object detection and tracking method based on key frame |
| WO2022042538A1 (en) * | 2020-08-24 | 2022-03-03 | 北京大学深圳研究生院 | Block-based point cloud geometric inter-frame prediction method and decoding method |
| CN113382242A (en) * | 2021-05-20 | 2021-09-10 | 北京大学深圳研究生院 | Code rate control method, device, equipment and computer readable storage medium |
| CN113573068A (en) * | 2021-07-28 | 2021-10-29 | 福州大学 | Improved V-PCC (V-PCC) inter-frame prediction method and system based on registration |
| CN114332259A (en) * | 2021-12-29 | 2022-04-12 | 福州大学 | Point cloud coding and decoding method based on vehicle-mounted laser radar |
| CN114663600A (en) * | 2022-04-18 | 2022-06-24 | 南京大学 | Point cloud reconstruction method and system based on self-encoder |
Non-Patent Citations (6)
| Title |
|---|
| "信息光学 光学信息处理、图像识别", 中国光学与应用光学文摘, no. 05, 25 October 2005 (2005-10-25) * |
| SHI-CHIUAN WANG等: "Dynamic detection technology for moving objects using 3D LiDAR information and RGB camera", 《2017 IEEE INTERNATIONAL CONFERENCE ON CONSUMER ELECTRONICS - TAIWAN (ICCE-TW)》, 27 July 2017 (2017-07-27) * |
| 常友渠;赵巧;粘永健;: "一种三维小波视频编码方法", 计算机工程与应用, no. 14, 11 May 2010 (2010-05-11) * |
| 徐鹏: "海量三维点云数据的组织与可视化研究", 《中国优秀硕士学位论文全文数据库》, no. 2, 15 February 2014 (2014-02-15) * |
| 郑宗安;: "基于AR技术与三维数字化点云重建解决方案的研究", 中国信息化, no. 11, 20 November 2019 (2019-11-20) * |
| 陈元相 等: "非一致性稀疏采样的LiDAR点云压缩方法", 《福州大学学报(自然科学版)》, vol. 49, no. 3, 30 June 2021 (2021-06-30) * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11836954B2 (en) | 3D point cloud compression system based on multi-scale structured dictionary learning | |
| US20230162402A1 (en) | Method and apparatus for processing a point cloud | |
| CN108171761B (en) | Point cloud intra-frame coding method and device based on Fourier image transformation | |
| US9819964B2 (en) | Limited error raster compression | |
| CN110443883B (en) | Plane three-dimensional reconstruction method for single color picture based on droplock | |
| Song et al. | Layer-wise geometry aggregation framework for lossless lidar point cloud compression | |
| CN109166160B (en) | Three-dimensional point cloud compression method adopting graph prediction | |
| CN113573068B (en) | Registration-based improved V-PCC inter-frame prediction method and system | |
| CN105409207A (en) | Feature-based image set compression | |
| CN113518226A (en) | G-PCC point cloud coding improvement method based on ground segmentation | |
| Sun et al. | A novel coding architecture for lidar point cloud sequence | |
| Sun et al. | A novel coding architecture for multi-line LiDAR point clouds based on clustering and convolutional LSTM network | |
| Tu et al. | Motion analysis and performance improved method for 3D LiDAR sensor data compression | |
| JP2024050705A (en) | Attribute information prediction method, encoder, decoder, and storage medium | |
| Wang et al. | R-pcc: A baseline for range image-based point cloud compression | |
| Fan et al. | Deep geometry post-processing for decompressed point clouds | |
| CN115474047A (en) | LiDAR point cloud encoding method and decoding method based on enhanced map correlation | |
| Wei et al. | Enhanced intra prediction scheme in point cloud attribute compression | |
| Sun et al. | A novel coding scheme for large-scale point cloud sequences based on clustering and registration | |
| CN114143556A (en) | Interframe coding and decoding method for compressing three-dimensional sonar point cloud data | |
| CN112995758B (en) | Encoding method, decoding method, storage medium and equipment for point cloud data | |
| Deng et al. | MASIC: Deep Mask Stereo Image Compression | |
| CN114554175B (en) | Classification rearrangement-based lossless compression method for two-dimensional point cloud distance images | |
| CN114915790A (en) | Two-dimensional regularization plane projection and coding and decoding method for large-scale point cloud | |
| CN111583295A (en) | Real-time dense optical flow computing method based on image block binarization Hash representation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |