JP2011216986A - Video transmission system, transmitting device, and repeating apparatus - Google Patents

Abstract

PROBLEM TO BE SOLVED: To make the transmission delay of video data small by being compared with such a technique for selecting packet groups of a new processing object by analyzing a sufficient amount of information about processed packet groups while suppressing a deterioration in the quality of video data transferred to a receiving device.SOLUTION: The receiving device 30 transmits a request including a file name being an identifier of video data to a video transmitting device 10. The video transmitting device 10 receives the request transmitted from a plurality of receiving devices 30, specifies an encoding condition, such as the maximum number of pixels optimal to the group of the receiving devices 30 and the number of hierarchies during encoding, and hierarchically encodes the video data according to the encoding condition. A video repeating apparatus 20 extracts a packet corresponding to a specific hierarchy or transfers all of the packets to a destination as they are in accordance with each of the receivers 30, for each receiving device 30 included in the plurality of receiving devices 30 when relaying a packet in the hierarchically encoded video data.

Description

  The present invention relates to a technique for transmitting hierarchically encoded video data.

  When transmitting video data from a transmission device that is a transmission source of video data to a plurality of reception devices such as a mobile phone, an SDTV (Standard Definition Television), or an HDTV (High Definition Television), for example, In some cases, video data of quality corresponding to the processing performance is transmitted. As this type of technology, video data to be transmitted is hierarchically encoded from high quality data to low quality data, and depending on the processing performance of the receiving device at the transmission destination, data of some or all layers is There is a technique called hierarchical encoding that is selectively transmitted. As a standard of hierarchical coding, ISO / IEC 14496-10 MPEG4 Advanced Video Coding, ITU-T H.264, etc. H.264 Annex G.

  A transmission apparatus or a relay apparatus that relays data transmission between the transmission apparatus and the reception apparatus selects a hierarchically encoded video data packet in units of the hierarchy according to the required quality, and the reception apparatus Transmission control is performed to transmit to In this case, at the time when a certain packet is processed in the transmission device or the relay device, in order to specify what the processing target packet has in the entire video data, information on the packet group that has been processed so far is stored. By performing analysis, prediction is performed for a subsequent packet group. Based on the prediction result, the transmission device or the relay device determines whether the packet to be processed should be transmitted or discarded without any problem (see, for example, Patent Document 1).

Republished publication WO2007 / 015482

However, in the technique described in Patent Document 1, since the analysis is performed after accumulating a sufficient amount of video data to perform the above analysis, for example, in the transmission path between the transmission device and the reception device. When relay devices are connected in multiple stages in cascade, the above-described analysis processing is performed many times in each relay device, and transmission delay of video data occurs depending on the number of cascade connection stages. There is a problem of end up.
Therefore, the present invention selects a new packet group to be processed by analyzing a sufficient amount of information about a processed packet group while suppressing a deterioration in the quality of video data delivered to the receiving apparatus. The purpose is to reduce the transmission delay of the video data as compared with a simple technique.

  In order to solve the above-described problem, the present invention provides a transmission device that transmits video data composed of a plurality of pictures that are hierarchically encoded in a spatial hierarchy and a temporal hierarchy to a reception device that is a destination of the video data; A relay device that relays the video data between the video transmission device and the reception device, wherein the transmission device converts a content of the video data that has been hierarchically encoded into a predetermined number of picture groups. Analysis means for analyzing in units of groups, and based on the analysis result by the analysis means, the video mode represented by the picture group included in each of the groups is a moving image mode, or the picture group Attribute information generating means for generating, for each of the groups, group attribute information indicating whether the motion of the video represented by is a still image mode; A transmission unit configured to encapsulate the layer-encoded video data and the generated group attribute information to generate a packet group and transmit the packet group to the reception device, and the relay device includes the packet group And when the group attribute information included in the packet group received by the receiving unit represents the video mode, a procedure for selecting packets with a weight on the time layer rather than the space layer And when the group attribute information represents the still image mode, a second function including a procedure for selecting packets with a weight in the space layer rather than the time layer And extracting means for extracting a packet to be transmitted from a packet group included in a group corresponding to each of the group attribute information, and by the extracting means To provide a video transmission system comprising: a transfer means for transferring destined to the reception device out packets.

  In a preferred aspect, the first function and the second function may include a procedure for extracting a packet according to performance related to video display in the receiving apparatus.

  In another preferable aspect, in the transmission device, the analysis means includes at least the presence or absence of a scene change represented by the video represented by the video data that has been hierarchically encoded, or the amount of movement of an object in the video. The attribute information generation means performs an analysis on either one and the definition of the video, and the attribute information generation means indicates that the analysis result by the analysis means that the scene has changed or the amount of motion of the object is greater than or equal to a threshold value. If it is the content, it is determined that the video mode is selected, and group attribute information indicating the video mode is generated, and the analysis result by the analysis means indicates that the scene has not changed or the object If the amount of motion is less than the threshold and the definition is greater than or equal to the threshold, it is determined that the still image mode is set, and It may generate a group attribute information indicating that the image mode.

  The present invention also relates to a transmission device that transmits video data composed of a plurality of pictures hierarchically encoded in a spatial hierarchy and a temporal hierarchy to a reception device that is a destination of the video data, and is hierarchically encoded. In addition, the content of the video data is represented by an analysis unit that analyzes a predetermined number of groups of pictures, and a group of pictures included in each group based on an analysis result by the analysis unit. Attribute information generating means for generating, for each of the groups, group attribute information indicating whether the video mode has a large video motion or a still image mode in which the video motion represented by the picture group is small. Generating a packet group by encapsulating the hierarchically encoded video data and the generated group attribute information; Further comprising a transmission means for transmitting addressed to provide a transmitting apparatus characterized by.

  Further, according to the present invention, receiving means for receiving a packet group transmitted from the transmitting device and group attribute information included in the packet group received by the receiving means are included in a group corresponding to the group attribute information. In the case of a moving image mode in which the motion of the video represented by the picture group is large, a first function including a procedure for selecting packets with a weight in the temporal layer rather than the spatial layer is used, and the group When the attribute information represents a still image mode in which the motion of the video represented by the group of pictures included in the group corresponding to the group attribute information is small, the packet is assigned with a weight in the space layer rather than the time layer. Using a second function including a selection procedure, a packet to be transmitted from a packet group included in a group corresponding to each of the group attribute information Extraction means for extracting comprising packets, a packet extracted by said extracting means, to provide a relay apparatus characterized by comprising a transfer means for transferring destined to the receiving apparatus that is the destination of the packet.

  According to the present invention, a new packet group to be processed is selected by analyzing a sufficient amount of information related to a processed packet group while suppressing deterioration in the quality of video data delivered to the receiving apparatus. The transmission delay of the video data can be reduced as compared with a simple technique.

It is a block diagram showing the structure of the video transmission system which concerns on embodiment of this invention. It is a block diagram showing the hardware constitutions of a video transmitter. It is a block diagram showing the function structure of a video transmitter. It is a figure showing the relationship between a hierarchy and video quality. It is a figure showing the structure of video data. It is a figure showing the relationship between the picture with a hierarchical structure, and a time-axis. It is a figure showing GOP attribute information. It is a figure showing multiplexing of video data. It is a block diagram showing the hardware constitutions of the video relay apparatus. It is a block diagram showing the functional structure of the video relay apparatus. It is a flowchart which shows operation | movement of an extraction part. It is a flowchart which shows operation | movement of an extraction part. It is a flowchart which shows operation | movement of an extraction part. It is a flowchart which shows operation | movement of an extraction part. It is a flowchart which shows operation | movement of an extraction part. It is a flowchart which shows operation | movement of an extraction part.

Hereinafter, an embodiment of the present invention will be described.
<Embodiment>
FIG. 1 is a block diagram showing a configuration of a video transmission system 1 according to an embodiment of the present invention. The video transmission system 1 includes a video transmission device 10, reception devices 30 </ b> A and 30 </ b> B (hereinafter referred to as reception device 30 when not distinguished from each other), and a transmission path between the video transmission device 10 and the reception device 30. Video relay devices 20A, 20B, and 20C (hereinafter referred to as video relay devices 20 when not distinguished from each other). Between the video transmission device 10 and the video relay device 20, between one video relay device 20 and another video relay device 20, and between the video relay device 20 and the reception device 30, a communication line or LAN (Local Area Network) ), The Internet, or a cable television network. Note that the number of video relay apparatuses 20 and reception apparatuses 30 for one video transmission apparatus 10 is not limited to the number shown in FIG.

  The video transmission device 10 is a computer such as a server device that stores video data and transmits it. The video transmitting apparatus 10 encapsulates video data after hierarchical encoding, and generates a plurality of packets having a predetermined data length. The video transmission device 10 transmits these packets to the reception device 30. The video relay device 20 is, for example, a router device, and is a device that relays video data while performing routing on the transmission path between the video transmission device 10 and the reception device 30. This video relay device 20 transfers packets to other video relay devices 20 or receives them while discarding the packets if necessary according to the processing performance of the receiving device 30 that is the destination of the video data. The packet is transferred to the device 30.

  The receiving device 30 is a device that receives video data and displays a video corresponding to the video data, such as a personal computer, an SDTV, an HDTV, or a mobile phone. The receiving device 30 group in FIG. In this way, receiving apparatuses having different processing performances are mixed. The receiving device 30 acquires video data by decapsulating the received packet, and displays video corresponding to the video data. In FIG. 1, the video relay apparatus 20 group has a two-stage cascade connection form, but the number of cascade connection stages is not limited to this. In addition, when the receiving device 30 is a device that does not support digital video data, such as an analog television, the video relay device 20 connected to the receiving device 30 has a conversion function for performing D / A conversion. Or a conversion device that performs D / A conversion between the reception device 30 and the video relay device 20 is provided.

  In this video transmission system 1, first, the reception device 30 transmits a request including a file name that is an identifier of video data to the video transmission device 10. Each video relay device 20 relays this request and transfers it to the video transmission device 10. The video transmission device 10 receives the request transmitted from the plurality of reception devices 30, determines what quality video should be provided to these reception devices 30, and based on the determination result Perform hierarchical coding. Here, the quality of the video refers to evaluation of the video such as the fineness of the video and the smoothness of the motion, and depends on the size of the number of pixels and the frame rate. High image quality means that, for example, the number of pixels of a video per picture is large or the frame rate is high, and low video quality means that, for example, the number of pixels of a video per picture is small or a frame is high. It means that the rate is small. The number of pixels is the total number of pixels when an image is displayed on the display surface of the receiving device 30, and is expressed by a numerical value such as 1024 × 768. The frame rate is an index representing how many times the video is updated per unit time.

  When determining how much quality video should be provided to the group of reception devices 30 by the video transmission device 10, for example, the communication performance of the reception device 30 and the performance of the video display function of the reception device 30. Various conditions such as the arithmetic processing capability of the receiving device 30, the power consumption of the receiving device 30, and the user settings of the receiving device 30 are taken into account, but in the following, based on the communication performance of the receiving device 30 As will be described with reference to an example of determination, the performance of this communication is ultimately related to a function related to the video display speed in the receiving device 30. The communication performance of the receiving device 30 is mainly the transmission rate of the communication network that accommodates the receiving device 30. For example, when the receiving device 30A is a mobile phone, transmission between the video relay device 20B and the receiving device 30A is performed. Since the path is configured by a mobile communication network, the transmission rate in the mobile communication network is an indicator of the communication performance of the receiving device 30A. For example, when the receiving device 30B is an HDTV, the transmission path between the video relay device 20C and the receiving device 30B is configured by an HDTV network, and therefore the transmission rate in the HDTV network is the communication performance of the receiving device 30B. It becomes an index. In response to requests from a plurality of receiving devices 30 having different conditions as described above, the video transmitting device 10 determines the optimum maximum number of pixels, the number of layers at the time of encoding, and the like for these receiving device 30 groups. An encoding condition is specified, and video data is hierarchically encoded in accordance with the encoding condition. When the video relay device 20 relays the packet of the hierarchically encoded video data, the video relay device 20 sends each of the reception devices 30 included in the plurality of reception devices 30 according to each of the reception devices 30. A packet corresponding to a specific layer is extracted, or all packets are transferred as they are to a destination without performing such extraction.

  Here, an outline of hierarchical encoding of video data performed by the video transmission device 10 will be described. Hierarchical encoding of video data means that video data is encoded in a layered state into video data that is of low quality but at least necessary for display and other video data that enhances the quality of the video at the time of display. It is to be. Here, consider an example of hierarchical coding having a three-layer structure of L0, L1, and L2. It is assumed that L0 is basic hierarchy data, L1 is first hierarchy data, and L2 is second hierarchy data. When only the basic layer data L0 is decoded, a low-quality video is obtained. When the basic layer data L0 and the first layer data L1 are decoded, a medium quality image is obtained, and when all of the L0, the first layer data L1, and the second layer data L2 are decoded, a high quality image is obtained. . At this time, the data of the base layer data L0 is data that is to be transmitted to the receiving device 30 without fail because it is the minimum required data in the receiving device 30 on the decoding side. On the other hand, the first layer data L1 and the second layer data L2 improve the quality of the video, but are not essential as the basic layer data L0. Therefore, the first layer data L1 and the second layer data L2 are transmitted to the receiving device 30 according to the required video quality. It is determined whether or not.

  FIG. 2 is a block diagram illustrating a hardware configuration of the video transmission device 10. The video transmitting apparatus 10 includes a CPU (Central Processing Unit) 11, a RAM (Random Access Memory) 12, a storage unit 13, and an input / output port 14, and these units are connected to each other via a bus. The CPU 11 reads out a program stored in the storage unit 13 and loads it into the RAM 12 and executes it, thereby controlling each unit of the video transmission device 10 and realizing various functions. The RAM 12 serves as a work area when the CPU 11 executes a program. The storage unit 13 is, for example, a large-capacity flash memory or HDD (Hard Disk Drive), and stores a program executed by the CPU 11 and video data to be transmitted. The input / output port 14 includes an input / output terminal connected to the transmission path, and transmits the video data packet and receives the request transmitted from the receiving device 30.

  FIG. 3 is a block diagram illustrating a functional configuration of the video transmission device 10. In the video transmission apparatus 10, the CPU 11 executes a program stored in the storage unit 13, whereby an input unit 101, a hierarchical encoding unit 102, an analysis unit 103, an attribute information generation unit 104, a multiplexing unit 105, and an output Each function of the unit 106 is realized. In the following description, each of these units will be described as the subject of operation, but the actual state of the subject is the CPU 11.

  Requests received by the input / output port 14 from the plurality of receiving devices 30 are input to the input unit 101. The input unit 101 supplies the request to the hierarchical encoding unit 102. The hierarchical encoding unit 102 refers to the file name of the video data included in the request, and reads the video data having this file name from the storage unit 13. Further, the hierarchical encoding unit 102 refers to the identifier or model code of the receiving device 30 included in the request, specifies the communication performance of the receiving device 30, and specifies the maximum pixel when encoding the video data. The number, the minimum number of pixels, and the number of layers are determined. Therefore, the storage unit 13 includes the identifier and model code of the receiving device 30 (attribute information of the receiving device 30) and the transmission rate (communication performance of the receiving device 30) in the communication network that accommodates the receiving device 30. Correspondingly stored in advance, the input unit 101 refers to it. The hierarchical encoding unit 102 hierarchically encodes the video data read from the storage unit 13 according to the maximum pixel number, the minimum pixel number, and the hierarchical number. The hierarchical encoding unit 102 outputs the hierarchically encoded video data to the analyzing unit 103.

  Here, FIG. 4 is a diagram showing the relationship between the hierarchy and the video quality. The types of layers in the hierarchical encoding of video data are roughly classified into a spatial layer and a time layer. The spatial hierarchy is a hierarchy of video data paying attention to the number of pixels. The time hierarchy is a hierarchy of video data paying attention to the frame rate. In the spatial hierarchy, the number of pixels increases as the hierarchy becomes higher, and the number of pixels decreases as the hierarchy becomes lower. In the time layer, the frame rate increases as the higher layer is reached, and the frame rate decreases as the lower layer is reached. The hierarchical encoding unit 102 hierarchically encodes video data in both a spatial hierarchy and a temporal hierarchy. The video data that has been hierarchically encoded by the hierarchical encoding unit 102 is in a state of being divided into spatial layers. Note that when the hierarchical encoding of the video data is performed by the hierarchical encoding unit 102, the number of layers in the space layer and the time layer is not necessarily the same, and the number of layers when encoding from the input unit 101 is different. May have different numbers of hierarchies.

  Next, FIG. 5 is a diagram showing the structure of video data. Hierarchically encoded video data is configured by unit data called GOP (Group Of Picture) being arranged in order. The GOP is a group of a predetermined number of frames, that is, pictures included in a certain time, for example, 32 pictures / second. As shown in FIG. 5, this GOP has a configuration of SSCEI, SPS (Sequence Parameter Set), PPS (Picture Parameter Set), Pic (Picture) 1, Pic 2, Pic 3,.

  SSCEI is information relating to the hierarchical structure in the GOP. SPS and PPS are initialization information required for performing initialization at the time of decoding, respectively. The SPS is information related to the entire GOP and includes, for example, a profile, the maximum number of pictures that can be referred to, and a video size. PPS is information relating to each picture, and includes, for example, the type of variable length coding, the initial value of the quantization step, the number of reference pictures, and the like. Pic is a picture. If the GOP is 32 pictures / second, the number of Pics in one GOP is 32. In FIG. 5, Pic is represented by assuming that one GOP is an N picture.

  Pics in GOP are classified into I picture (Intra Picture), P picture (Predictive Picture), and B picture (Bi-directionally predictive Picture) depending on the predictive coding method. An I picture (Intra Picture) is a picture obtained by encoding only that picture independently. That is, all macroblock types in an I picture are intraframe coding. An I picture maintains GOP independence, and one I picture exists at the head of one GOP. Since this I picture is a key frame in the GOP, it will be called a key frame in the following description, and the other P pictures and B pictures will be called non-key frames. A P picture is a picture obtained by forward predictive coding between pictures. Macroblock types in P pictures include intra-frame coding and forward inter-frame predictive coding. The B picture is a picture obtained by predictive coding from the past and the future. Macroblock types in B pictures include intra-frame coding and forward, backward and bidirectional inter-frame predictive coding.

  FIG. 6 is a diagram illustrating a relationship between a picture having a hierarchical structure and a time axis. Here, a case where video data is hierarchically encoded in three layers in both the space layer and the time layer is illustrated. The horizontal axis represents time t, and it is assumed that time has passed toward the right. The number given to Pic is the order of each picture in time series. The alphabet attached to each Pic represents an I picture, a B picture, or a P picture. TID (Temporal ID) is information for identifying each layer in the time layer, and DID (Dependency ID) is information for identifying each layer in the spatial layer. For both TID and DID, the number attached represents the number of layers, “0” is the lower layer, “1” is the middle layer, and “2” is the upper layer. As shown in FIG. 6, each picture at each time t has three layers of data with DIDs “0”, “1”, and “2”. In addition, the same TID is assigned to a picture composed of data having a plurality of DIDs at each time t. Each picture at time t has a reference relationship represented by an arrow in FIG.

  For example, when low-quality video data is transmitted in the time hierarchy, the video relay device 20 extracts only packets based on the picture with the TID “0” from the received packet group and addresses the packets to the reception device 30. Forward. In this case, the receiving device 30 displays an image corresponding to pictures corresponding to t0 and t4 in FIG. In other words, since the frame rate is small, the video is not very moving like so-called frame advance, and the video quality is low. When transmitting medium quality video data in the time hierarchy, the video relay apparatus 20 extracts and receives only packets based on pictures with TIDs “0” and “1” from the received packet group. Transfer to device 30. Furthermore, when high-quality video data is transmitted in the time hierarchy, the video relay device 20 transfers all received packet groups to the receiving device 30. In this case, since all the pictures with TIDs “0”, “1”, and “2” corresponding to each time t in FIG. 6 are prepared, the frame rate becomes high, the motion is smooth, and the video quality is improved. Is expensive.

  As described above, the video quality of the video data to be transmitted and the transmission target data have been described using the TID, but the same idea applies to the DID. That is, the video relay device 20 extracts a packet based on the data of DID “0” and transfers it to the receiving device 30 when transmitting low-quality video data in the spatial hierarchy. In this case, the receiving device 30 displays a coarse video with low definition. Also, the video relay device 20 extracts packets based on the data of the DID “0” and “1” and transfers them to the receiving device 30 when transmitting medium quality video data in the spatial hierarchy. Further, the video relay device 20 transfers all packets to the receiving device 30 when transmitting high-quality video data in the spatial hierarchy. In this case, the receiving device 30 displays a high-definition video.

  Returning to the description of FIG. The analysis unit 103 performs the following analysis on the hierarchically encoded video data from the three types of viewpoints of scene change estimation, motion amount estimation, and definition estimation for each GOP. The analysis unit 103 estimates that a scene change has occurred when the number of intra macroblock absolute values in an I picture included in one GOP exceeds a threshold, and determines that the scene change has occurred in the I picture included in one GOP. When the number of intra macroblock absolute values does not exceed the threshold, it is estimated that no scene change has been performed. A scene change is a change in a scene represented by a video, for example, an angle changes. Further, when a motion vector having a certain threshold value or more exists in a plurality of adjacent macroblocks, the analysis unit 103 detects the plurality of adjacent macroblocks as one object, and the number of detected objects Is specified as an index representing the amount of motion in the GOP. In addition, the analysis unit 103 identifies the relative variation rate of the number of predictive coding macroblocks between non-key frames as an index representing the amount of motion in the GOP. The index indicating the amount of movement may be only the former or the latter, or the analysis unit 103 may specify both of them. Furthermore, in the intra-frame coding and the inter-frame prediction coding, the analysis unit 103 specifies the ratio of the area rounded after quantization with respect to the entire picture area as an index representing the definition. The analysis unit 103 supplies the analysis result for each GOP to the attribute information generation unit 104 together with the hierarchically encoded video data.

  The attribute information generation unit 104 generates GOP attribute information representing the attribute of the picture included in the GOP for each GOP from the analysis result for each GOP and the hierarchically encoded video data. FIG. 7 shows the contents of GOP attribute information. The GOP attribute information includes GOP video characteristics, the number of pictures in the GOP, the total data size of all pictures in the GOP, a data size total list for each picture, and a data size total list for each DID. Among these, the video characteristics of the GOP represent the characteristics of the video in the GOP. Based on the analysis result for each GOP, the attribute information generation unit 104 determines whether the video characteristic of the GOP is a video with a large motion (moving image mode) or a video with a small motion (still image mode). Classify.

  The video characteristics of the GOP include either a moving image mode which means that the motion in the video is large or a still image mode which means that the motion in the video is small (including no motion). The video relay apparatus 20 uses this video specification when performing the extraction process for extracting the packet as described above. If the attribute information generation unit 104 is an estimation result of scene change in the analysis result for each GOP, or if the index value indicating the amount of motion is greater than or equal to a certain value, the video characteristics of the GOP are in the moving image mode. Judge. Further, the attribute information generation unit 104 is an estimation result indicating that no scene change has been performed in the analysis result for each GOP, or an index value indicating the amount of motion is less than a threshold and an index indicating the definition If the value is greater than or equal to the threshold, it is determined that the video characteristics of the GOP are still image modes.

  Further, the attribute information generation unit 104 analyzes the hierarchically encoded video data for each GOP, and the number of pictures in the GOP, the total data size of all the pictures in the GOP, the data size total list of each picture, and the data size for each DID A total list is generated as GOP attribute information. The number of pictures in a GOP is the number of pictures included in one GOP. The total data size of all pictures in a GOP is the sum of data sizes including all layers of all pictures included in one GOP. The total data size list of each picture is a list of the total data size of one picture made up of a set of data having a plurality of different DIDs for each picture. The data size total list for each DID is a list of the total data sizes for each DID for each DID. The attribute information generation unit 104 outputs the generated GOP attribute information to the multiplexing unit 105 together with the hierarchically encoded video data.

  The multiplexing unit 105 multiplexes the video data encoded for each layer and adds GOP attribute information to the head thereof. Here, the multiplexing of the video data means that the encoded video data divided for each of the spatial hierarchies are put together in a sequence on the time axis.

  FIG. 8 is a diagram showing multiplexing of video data. Here, a case where video data is hierarchically encoded in three layers in both the space layer and the time layer is illustrated. In FIG. 8, “GopInfo” represents the GOP attribute information described above. Further, SSCEI, SPS, PPS and Pic have the same contents as in FIG. A slice is a unit of data constituting hierarchical video data. The numbers given to SPS, PPS, TID, and DID represent the number of layers to which they belong. When the video data is input to the multiplexing unit 105, slices are arranged in order along the time axis in a state of being divided for each DID hierarchy. The multiplexing unit 105 multiplexes the video data divided for each DID layer, adds GOP attribute information to the head, and supplies the GOP attribute information to the output unit 106.

  The output unit 106 functions as a transmission unit that transmits a packet group of video data. The output unit 106 encapsulates video data that has been hierarchically encoded and has GOP attribute information added thereto, and thereby a plurality of packets having a predetermined data length. Divide into The output unit 106 sequentially outputs these packets to the transmission path through the input / output port 14 with the receiving device 30 as a destination.

Next, the video relay device 20 will be described.
FIG. 9 is a block diagram showing the hardware configuration of the video relay apparatus 20. The video relay apparatus 20 includes a CPU (Central Processing Unit) 21, a volatile storage unit 22, a non-volatile storage unit 23, an input port 24, and an output port 25. These units are connected to each other via a bus. Yes. The CPU 21 reads out a program stored in the nonvolatile storage unit 23, loads it into the volatile storage unit 22, and executes it, thereby controlling each unit of the video relay device 20 and realizing various functions. The volatile storage unit 22 becomes a work area when the CPU 21 executes the program. The nonvolatile storage unit 23 is, for example, a flash memory, and stores a program relating to each function of the video relay device 20 realized by the CPU 21 and a routing table for performing packet routing. The input port 24 includes an input terminal connected to the transmission path, and receives video data packets. The output port 25 includes an output terminal connected to the transmission path, and outputs a packet to the transmission path.

  FIG. 10 is a block diagram illustrating a functional configuration of the video relay apparatus 20. The video relay device 20 implements the functions of the input unit 201, the extraction unit 202, and the output unit 204 by the CPU 21 reading and executing a program stored in the nonvolatile storage unit 23. In the following description, each of these units will be described as the subject of operation, but the actual state of the subject is the CPU 21. The input unit 201 is a receiving unit that receives a packet group of video data, receives a packet from a transmission path connected through the input port 24, and supplies the packet to the extraction unit 202. The extracting unit 202 extracts only packets to be transmitted by substituting these packets into a predetermined function, and discards the others. In the following description, this function is called a cost function and is stored in the nonvolatile storage unit 23.

  The extraction unit 202 supplies the extracted packet to the output unit 204. The output unit 204 is a transfer unit that transfers the extracted packet to the receiving device 30, and transmits the packet to the video relay device 20 or the receiving device 30 existing on the subsequent transmission path. Here, the number of packet streams supplied by the extraction unit 202 is not limited to one, but a plurality of reception devices 30 connected to the video relay device 20 receive video data corresponding to a plurality of video qualities in response to requests. Packets can be supplied. For example, on a subsequent transmission path of a certain video relay device 20, a mobile phone and an HDTV exist as the receiving device 30, the receiving device 30 of the mobile phone requests low-quality video data, and the HDTV receiving device 30 Assume that high-quality video data is requested. In this case, the extraction unit 202 extracts, as a transmission target, packets corresponding to all hierarchies of the hierarchically encoded data for the HDTV receiving device 30, and the hierarchical encoding for the mobile phone receiving device 30. A packet corresponding to a lower hierarchy of data is extracted as a transmission target.

  Next, how the extraction unit 202 performs packet extraction using the cost function and GOP attribute information will be described. The content of the cost function differs depending on whether the video characteristic of the GOP in the GOP attribute information is the still image mode or the moving image mode. When the video characteristic of the GOP is the still image mode, the cost function extracts packets with an emphasis on the space layer rather than the time layer. That is, this cost function is a first function including a procedure for selecting packets with a weight in the time layer rather than the space layer. On the other hand, when the video characteristic of the GOP is the moving image mode, the cost function extracts packets with an emphasis on the time layer rather than the space layer. That is, this cost function is a second function including a procedure for selecting packets with a weight on the space layer rather than the time layer.

Here, with reference to FIGS. 11 to 16, the packet extraction operation by the extraction unit 202 will be described in the case where the video characteristics of the GOP are the still image mode and the moving image mode. In the following description, each content (FIG. 7) included in the GOP attribute information is expressed as follows.
PicProp // Video characteristics of GOP
GopPicNum // Number of pictures in GOP
Gop # PicTotalSize // Total data size of all pictures in GOP
PicSize [GopPicNum] // Data size total list for each picture
GOP # DidSize [MaxDID + 1] // Data size total list for each DID MaxTID is the maximum value of the time hierarchy when the lowest hierarchy is 0, and MaxDIDD is 0 for the lowest hierarchy Is the maximum value of the spatial hierarchy.

First, an operation when the extraction unit 202 extracts a packet to be transmitted using a cost function when the video characteristic of the GOP is the still image mode will be described.
First, the extraction unit 202 obtains a TargetBufSize by the following equation, with the target transmission rate in the transmission path to which the video relay apparatus 20 is connected as TargetBitRate. In FIG. 1, for example, when the receiving device 30A is a mobile phone, the transmission path between the video relay device 20B and the receiving device 30A is configured by a mobile communication network, so the video relay device 20B is determined in advance. The target transmission rate of the mobile communication network or the target transmission rate of the mobile communication network measured periodically or as needed is set as TargetBitRate. Similarly, for example, when the receiving device 30B is an HDTV, the transmission path between the video relay device 20C and the receiving device 30B is configured by an HDTV network, and thus the video relay device 20C is connected to a predetermined HDTV network. The target transmission rate or the target transmission rate of the HDTV network measured periodically or as needed is set as TargetBitRate.
TargetBufSize = TargetBitRate / GopPicNum × FrameRate
Note that FrameRate is a frame rate related to the video quality described above.

なお、tid( i )は、先頭から数えてｉ番目のピクチャが属するＴＩＤである。 Next, the extraction unit 202 obtains a function subtotalGopTidSize (TID) by the following equation. The function subtotalGopTidSize (TID) means the total data amount of pictures belonging to the TID of the argument hierarchy.

Note that tid (i) is a TID to which the i-th picture from the top belongs.

Next, the extraction unit 202 performs the process shown in FIG. 11 to determine to which TID layer transmission is possible, that is, the number of time layers that can be transmitted.
In FIG. 11, the extraction unit 202 initializes TotalTidGop, which is the sum of the total data size of pictures for each TID, to set TotalTidGop = 0 (step S1), and further initializes id to id = 0 (step S2). ). id is a value of each time layer when the lowest layer is 0, and the maximum value is the maximum value maxTID of the time layer.
Next, the extraction unit 202 obtains a value of tmp by a formula of tmp = TotalTidGop + subtotalGopTidSize (id) (Step S3), where totalTidGop means the sum of data sizes up to NewMaxTID-1. Here, when tmp is equal to or smaller than TargetBufSize (step S4; No), the extraction unit 202 sets totalTidGop = tmp (step S5). Next, the extraction unit 202 increments id by one (step S6), and determines whether the id is equal to or less than the maximum value maxTID of the time hierarchy (step S7). If id is less than or equal to the maximum value maxTID (step S7; Yes), the processing of the extraction unit 202 returns to step S3 again. The extraction unit 202 repeats the processing from step S3 to step S7 until id exceeds the maximum value maxTID of the time hierarchy, sets id at that time as NewMaxTID (step S8), and ends the processing illustrated in FIG. Further, in the process of performing steps S3 to S7, when the tmp exceeds TargetBufSize (step S4; Yes), the extraction unit 202 sets the id at that time as NewMaxTID (step S8). The processing shown in FIG.

  As a result of the above processing, for example, when NewMaxTID = 0, it means that only some pictures in the lowest layer can be transmitted, and when NewMaxTID = maxTID + 1, for example, This means that pictures in the time hierarchy can be transmitted. Further, if MaxTID> = NewMaxTID> 0, it means that there is a high possibility that some, if not all, pictures having the NewMaxTID time hierarchy can be transmitted. Therefore, when maxTID> = NewMaxTID> 0, the extraction unit 202 selects a picture that can be transmitted from a group of pictures having a time hierarchy of NewMaxTID.

In general, when the picture encoding method is the same, the larger the data amount, the larger the content of the video compared to the context, or the more complicated the video. Yes. Since such video greatly affects the video quality, it is desirable that the video is preferentially transmitted from a picture with a large amount of data. Therefore, the extraction unit 202 sorts the pictures having the NewMaxTID time hierarchy in the order of the data amount (PicSize [i]) of each picture. Then, the extraction unit 202 stores the sorted result in a table represented by the following formula.
N = GopPicNum + (2 ^ (MaxTID-NewMaxTID)-1) / (2 ^ (MaxTID-NewMaxTID))
The order is an index value when counting from the larger PicSize on this table, and the minimum value is 0 and the maximum value is N. ListPicSizeOrder [order] .PicNum is the order of pictures counted from the top in the GOP, and ListPicSizeOrder [order] .PicSize is the data size of the corresponding picture.

Next, the extraction unit 202 performs processing shown in FIG. 12 to select a picture that can be transmitted from pictures having a time hierarchy of NewMaxTID.
In FIG. 12, the extraction unit 202 first initializes subtotalGopTidSize (NewMaxTID), which means the total data size of pictures whose time hierarchy is NewMaxTID, and sets subtotalGopTidSize (NewMaxTID) = 0 (step S11). Is set to order = 0 (step S12). In the above table, order is the index value when counting from the larger PicSize, where the minimum value is 0 and the maximum value is N.

  Next, the extraction unit 202 obtains a value of tmp according to a mathematical formula of tmp = totalTidGop + ListPicSizeOrder [order] .PicSize (step S13). Here, when tmp is equal to or smaller than TargetBufSize (step 14; No), the extraction unit 202 sets subtotalGopTidSize (NewMaxTID) = subtotalGopTidSize (NewMaxTID) + ListPicSizeOrder [order] .PicSize (step S15). Next, the extraction unit 202 sets totalTidGop = tmp (step S16), increments order by 1 (step S17), and determines whether the order is less than the maximum value N (step S18). If orderd is less than the maximum value N (step S18; Yes), the process of the extraction unit 202 returns to step S13 again. The extraction unit 202 repeats the processing from step S13 to step S18 until order reaches the maximum value N or more, determines the order value at that time as orderNum (step S19), and ends the processing shown in FIG. . Further, in the course of performing the processing from step S13 to step S18, when the tmp exceeds TargetBufSize (step S14; Yes), the extraction unit 202 determines the order value at that time as orderNum (step S19). ), The process shown in FIG.

When the extraction unit 202 selects a picture that can be transmitted from the pictures having the NewMaxTID time hierarchy as described above, the extraction unit 202 performs a process illustrated in FIG. 13 to create a list of pictures to be transmitted from the GOP. .
First, the extraction unit 202 initializes i to i = 0 (step S21). The minimum value of i is 0 and the maximum value is GopPicNum. Next, the extraction unit 202 determines whether tid (i) is less than NewMaxTID (step S22). If tid (i) is less than NewMaxTID (step S22; Yes), the extraction unit 202 sets SendingPictureList [i] = tid (i) (step S23). SendingPictureList [i] is a list of pictures to be transmitted among GOPs. Next, the extraction unit 202 increments i by 1 (step S25), confirms that i is less than GopPicNum (step S26; Yes), and returns to the process of step S22. In step S22, when the extraction unit 202 determines that tid (i) is not less than NewMaxTID (step S22; No), it sets SendingPictureList [i] = − 1 (step S24).

  In step S26, when i is equal to or greater than GopPicNum (step S26; No), the extraction unit 202 determines whether or not the condition 0 <NewMaxTID && NewMaxTID <= MaxTID is satisfied (step S27). When this condition is satisfied (step S27; Yes), the extraction unit 202 sets order = 0 (step S28). Next, the extraction unit 202 sets SendingPictureList [ListPicSizeOrder [order] .PicNum] = NewMaxTID (step S29). The extraction unit 202 increments the order by 1 (step S30), and then determines whether the order is less than orderNum (step S31). If order is less than orderNum (step S31; Yes), the process of the extraction unit 202 returns to step S29, and the processes of steps S29 to S31 are repeated until order becomes equal to or greater than orderNum. In step S27, if the condition 0 <NewMaxTID && NewMaxTID <= MaxTID is not satisfied (step S27; No), the processing of the extraction unit 202 ends.

  Then, the extraction unit 202 extracts and transmits a packet corresponding to this picture, with a picture whose array value in the list is not −1 (SendingPictureList [i] = − 1) as a transmission target. On the other hand, the extraction unit 202 discards a picture whose array value in the list is −1 (SendingPictureList [i] = − 1) without setting it as a transmission target. As a result, some pictures in the time hierarchy are discarded.

なお、did( i )は、先頭から数えてｉ番目のピクチャが属するDIDである。 Next, an operation when the extraction unit 202 extracts a packet using a cost function when the video characteristic of the GOP is the moving image mode will be described.
The extraction unit 202 obtains the function subtotalGopDidSize (DID) by the following equation. The function subtotalGopDidSize (DID) means the total data amount of pictures belonging to the DID of the argument hierarchy.

Here, did (i) is a DID to which the i-th picture from the beginning belongs.

Next, the extraction unit 202 performs the processing shown in FIG. 14 to determine to which DID hierarchy transmission is possible, that is, the number of time hierarchy layers that can be transmitted.
In FIG. 14, the extraction unit 202 initializes TotalDidGop, which is an integration of the total data size of pictures for each DID, to set TotalDidGop = 0 (step S41), and also initializes id to id = 0 (step S42). ). id is a value of each time layer when the lowest layer is 0, and the maximum value is the maximum value maxDID of the time layer.

  Next, the extraction unit 202 obtains a value of tmp by a formula of tmp = TotalDidGop + subtotalGopDidSize (id) (Step S43), where totalDidGop means the sum of data sizes up to NewMaxDID-1. Here, when tmp is equal to or smaller than TargetBufSize (step S44; No), the extraction unit 202 sets totalDidGop = tmp (step S45). Next, the extraction unit 202 increments id by one (step S46), and determines whether the id is equal to or less than the maximum value maxDID of the time hierarchy (step S47). If id is not more than the maximum value maxDID (step S47; Yes), the processing of the extraction unit 202 returns to step S43 again. The extraction unit 202 repeats the processing from step S43 to step S47 until id exceeds the maximum value maxDID of the time hierarchy, sets id at that time as NewMaxDID (step S48), and ends the processing illustrated in FIG. Further, in the course of performing the processing from step S43 to step S47, the extraction unit 202 sets id at that time as NewMaxDID (step S48) when tmp exceeds TargetBufSize (step S44; FIG. 14). The processing shown in FIG.

  As a result of the above processing, for example, when NewMaxDID = 0, it means that only some pictures of the lowest layer can be transmitted, and when NewMaxDID = maxDID + 1, for example, This means that pictures in the time hierarchy can be transmitted. When MaxDID> = NewMaxDID> 0, it is highly possible that some, if not all, pictures having the NewMaxDID time hierarchy can be transmitted. Therefore, when maxDID> = NewMaxDID> 0, the extraction unit 202 selects a picture that can be transmitted from a picture group having a time hierarchy of NewMaxDID.

The extraction unit 202 sorts the pictures having the NewMaxDID time hierarchy in the order of the data amount (PicSize [i]) of each picture. Then, the extraction unit 202 stores the sorted result in a table represented by the following formula.
N = GopPicNum + (2 ^ (MaxDID-NewMaxDID)-1) / (2 ^ (MaxDID-NewMaxDID))
However, order is an index value when counting from the larger PicSize on this table, and the minimum value is 0 and the maximum value is N. Further, ListPicSizeOrder [order] .PicNum is the order of pictures counted from the top in the GOP, and ListPicSizeOrder [order] .PicSize is the data size of the corresponding picture.

Next, the extraction unit 202 performs processing shown in FIG. 15 to select a picture that can be transmitted from pictures having a time hierarchy of NewMaxDID.
In FIG. 15, the extraction unit 202 first initializes subtotalGopDidSize (NewMaxDID), which means the sum of the data sizes of pictures whose time hierarchy is NewMaxDID, and sets subtotalGopDidSize (NewMaxDID) = 0 (step S51). Is set to order = 0 (step S52). In the above table, order is the index value when counting from the larger PicSize, where the minimum value is 0 and the maximum value is N.

  Next, the extraction unit 202 obtains a value of tmp according to a mathematical formula of tmp = totalDidGop + ListPicSizeOrder [order] .PicSize (step S53). Here, when tmp is equal to or smaller than TargetBufSize (step 54; No), the extraction unit 202 sets subtotalGopDidSize (NewMaxDID) = subtotalGopDidSize (NewMaxDID) + ListPicSizeOrder [order] .PicSize (step S55). Next, the extraction unit 202 sets totalDidGop = tmp (step S56), increments order by 1 (step S57), and determines whether the order is less than the maximum value N (step S58). If orderd is less than the maximum value N (step S58; Yes), the process of the extraction unit 202 returns to step S53 again. The extraction unit 202 repeats the processing from step S53 to step S58 until order reaches the maximum value N or more, determines the value of order at that time as orderNum (step S59), and ends the processing shown in FIG. . Further, in the process of performing steps S53 to S58, when the tmp exceeds TargetBufSize (step S54; Yes), the extraction unit 202 determines the order value at that time as orderNum (step S59). ), The process shown in FIG.

When the extraction unit 202 selects a picture that can be transmitted from the pictures having the NewMaxDID time hierarchy as described above, the extraction unit 202 performs a process illustrated in FIG. 16 to create a list of pictures to be transmitted from the GOP. .
First, the extraction unit 202 initializes i to i = 0 (step S61). The minimum value of i is 0 and the maximum value is GopPicNum. Next, the extraction unit 202 determines whether did (i) is less than NewMaxDID (step S62). If did (i) is less than NewMaxDID (step S62; Yes), the extraction unit 202 sets SendingPictureList [i] = did (i) (step S63). SendingPictureList [i] is a list of pictures to be transmitted among GOPs. Next, the extraction unit 202 increments i by 1 (step S65), confirms that i is less than GopPicNum (step S66; Yes), and returns to the process of step S62. In step S62, when the extraction unit 202 determines that did (i) is not less than NewMaxDID (step S62; No), it sets SendingPictureList [i] = − 1 (step S64).

  In step S66, when i is equal to or greater than GopPicNum (step S66; No), the extraction unit 202 determines whether the condition 0 <NewMaxDID && NewMaxDID <= MaxDID is satisfied (step S67). When this condition is satisfied (step S67; Yes), the extraction unit 202 sets order = 0 (step S68). Next, the extraction unit 202 sets SendingPictureList [ListPicSizeOrder [order] .PicNum] = NewMaxDID (step S69). The extraction unit 202 increments the order by one (step S70), and then determines whether the order is less than orderNum (step S71). If order is less than orderNum (step S71; Yes), the process of the extraction unit 202 returns to step S69, and the processes of steps S69 to S71 are repeated until order is equal to or greater than orderNum. In step S67, when the condition 0 <NewMaxDID && NewMaxDID <= MaxDID is not satisfied (step S67; No), the processing of the extraction unit 202 ends.

  Then, the extraction unit 202 extracts and transmits a packet corresponding to this picture, with a picture whose array value in the list is not −1 (SendingPictureList [i] = − 1) as a transmission target. On the other hand, the extraction unit 202 discards a picture whose array value in the list is −1 (SendingPictureList [i] = − 1) without setting it as a transmission target. As a result, some pictures in the time hierarchy are discarded.

  Thus, according to the present embodiment, the video transmitting apparatus 10 adds GOP attribute information generated by analyzing the hierarchically encoded video data to the head of the GOP. Then, when the video relay apparatus 20 performs packet extraction, it is only necessary to substitute GOP attribute information into the cost function. For example, in order to specify what meaning the processing target packet has in the entire video data Compared with the case of analyzing information on a packet group that has been processed so far, the time required for packet extraction processing is shortened. Therefore, in comparison with a technique for selecting a new packet group to be processed by analyzing a sufficient amount of information about the processed packet group, the video relay apparatus 20 is connected in multiple stages. However, it is possible to suppress a delay related to data transmission. Also, when the video characteristics of the GOP included in the GOP attribute information are in the moving picture mode, the video relay apparatus 20 extracts packets with an emphasis on the spatial hierarchy, and the video characteristics of the GOP are in the still picture mode. First, packets are extracted with an emphasis on the time hierarchy. As described above, the video relay apparatus 20 discards a packet having a small influence on the quality of the video in accordance with the characteristics of the video data for each GOP, and therefore discards the packet without considering such an influence. In comparison, the video quality of the video data delivered to the receiving apparatus can be kept high.

<Modification>
The above embodiment may be modified as follows.
<Modification 1>
In the above embodiment, when the video transmitting apparatus 10 determines what quality video should be provided to the receiving apparatus 30 group, the determination is made based on the communication performance of the receiving apparatus 30. In addition to this, as a material for this determination, the communication performance of the receiving device 30, the performance of the video display function of the receiving device 30, the arithmetic processing capability of the receiving device 30, the power consumption of the receiving device 30, the receiving device There are 30 user settings. In this case, the information used as the material for the determination and the video quality level determined by the receiving device 30 itself based on the material are notified from the receiving device 30 to the video transmitting device 10, and the video transmitting device 10 You will know the contents.

<Modification 2>
In the above-described embodiment, the attribute information generation unit 104 determines whether the video characteristics of the GOP is a video with a large motion (moving image mode) based on the analysis result for each GOP, or a video with a high sense of detail and a small motion. (Still image mode) was determined, but an intermediate mode that does not belong to any of these modes is considered, and it may be determined that any of these three modes is applicable. Good. Such an intermediate mode cost function extracts packets by treating both the time layer and the space layer equally.
In the embodiment, the analysis unit 103 uses the analysis results of both the scene change estimation and the motion amount estimation as the condition corresponding to the moving image mode. However, at least one of them is used. Also good.

<Modification 3>
In the above-described embodiment, the video transmission device 10 stores video data in the storage unit 13 provided therein, and reads video data to be transmitted from the video data. However, the present invention is not limited to this. The video transmission device 10 may acquire video data from a storage device such as an external HDD or a digital video camera connected to itself or an imaging device.

<Modification 4>
In the above-described embodiment, in the video transmission apparatus 10, the hierarchical encoding unit 102 and the analysis unit 103 have separate functions, but these are combined to form a hierarchical encoding analysis unit, which performs hierarchical encoding of video data. At the same time, analysis processing may be performed. In this way, when outputting the analysis result for each GOP, the hierarchical coding analysis unit estimates that a scene change has been performed when the accumulation of absolute values of inter-frame differences exceeds a set threshold value. It becomes possible. Further, when the analysis result for each GOP is output, the hierarchical coding analysis unit occupies the code of the high-frequency area before quantization for the whole picture area in intra-frame coding and inter-frame prediction coding. It is also possible to calculate the ratio as an index representing the definition.

<Modification 5>
In the above-described embodiment, the video transmission device 10 includes the hierarchical encoding unit 102, which performs hierarchical encoding of video data. However, the video data to be transmitted has been previously hierarchically encoded. In this case, the hierarchical encoding unit 102 may output the input hierarchically encoded video data to the analyzing unit 103 as it is. In addition, when the video transmission device 10 is connected to a separate hierarchical encoder, and the video data to be transmitted by the video transmission device 10 is limited to the one that has been previously hierarchically encoded, The transmission apparatus 10 does not have to implement the function of the hierarchical encoding unit 102.

<Modification 6>
The program executed in the video transmission device 10 and the video relay device 20 is a computer-readable recording such as a magnetic recording medium (magnetic tape, magnetic disk, etc.), an optical recording medium (optical disc, etc.), a magneto-optical recording medium, and a semiconductor memory. It can be provided to the video transmission device 10 and the video relay device 20 in a state stored in a medium. In this case, an interface for reading a program from the recording medium may be provided in the video transmission device 10 and the video relay device 20. It is also possible to cause the video transmission device 10 and the video relay device 20 to download the program via a network.

DESCRIPTION OF SYMBOLS 1 ... Video transmission system, 10 ... Video transmission apparatus, 11, 21 ... CPU, 12 ... RAM, 13 ... Memory | storage part, 14 ... Input / output port, 25 ... Output port, 20, 20A, 20B, 20C ... Video relay apparatus, DESCRIPTION OF SYMBOLS 22 ... Volatile memory | storage part, 23 ... Non-volatile memory | storage part, 24 ... Input port, 30, 30A, 30B ... Receiver, 101, 201 ... Input part, 102 ... Hierarchical encoding part, 103 ... Analysis part, 104 ... Attribute Information generation unit, 105 ... multiplexing unit, 106, 204 ... output unit, 202 ... extraction unit.

Claims (5)

A transmitting device that transmits video data composed of a plurality of pictures that are hierarchically encoded in the spatial layer and the temporal layer, to a receiving device that is a destination of the video data;
A relay device that relays the video data between the video transmission device and the reception device;
The transmitter is
Analyzing means for analyzing the content of the hierarchically encoded video data in units of a group consisting of a predetermined number of pictures;
Based on the analysis result of the analysis means, the still image is a moving image mode in which the motion of the video represented by the group of pictures included in each of the groups is large or the motion of the video represented by the group of pictures is small. Attribute information generating means for generating group attribute information indicating whether the mode is set for each of the groups;
A transmission unit that encapsulates the hierarchically encoded video data and the generated group attribute information to generate a packet group, and transmits the packet group to the receiving device;
The relay device is
Receiving means for receiving the packet group;
When the group attribute information included in the packet group received by the receiving unit represents the moving image mode, a first function including a procedure for selecting packets with a weight on the time layer rather than the space layer is provided. And when the group attribute information represents the still image mode, a second function including a procedure for selecting packets with a weight in the space layer rather than the time layer is used, and each of the groups is used. Extraction means for extracting a packet to be transmitted from a packet group included in a group corresponding to the attribute information;
A video transmission system comprising: a transfer unit configured to transfer the packet extracted by the extraction unit to a receiving device. The video transmission system according to claim 1, wherein the first function and the second function include a procedure of extracting a packet according to performance related to video display in the reception device. In the transmitter,
The analysis means is related to at least one of the presence / absence of a scene change represented by the video represented by the video encoded by the hierarchical encoding or the amount of movement of an object in the video, and the definition of the video. Do the analysis,
The attribute information generation unit determines that the video mode is selected when the analysis result by the analysis unit includes a change in the scene or the amount of motion of the object is greater than or equal to a threshold value. Group attribute information indicating that the mode is selected, and the analysis result by the analysis means is that the scene does not change or the amount of movement of the object is less than a threshold and the definition is greater than or equal to the threshold. 3. The video transmission system according to claim 1, wherein in some cases, it is determined that the still image mode is selected, and group attribute information indicating the still image mode is generated. A transmission device that transmits video data composed of a plurality of pictures that are hierarchically encoded in a spatial hierarchy and a temporal hierarchy to a reception device that is a destination of the video data,
Analyzing means for analyzing the content of the hierarchically encoded video data in units of a group consisting of a predetermined number of pictures;
Based on the analysis result of the analysis means, the still image is a moving image mode in which the motion of the video represented by the group of pictures included in each of the groups is large or the motion of the video represented by the group of pictures is small. Attribute information generating means for generating group attribute information indicating whether the mode is set for each of the groups;
A transmission apparatus comprising: transmission means for encapsulating the hierarchically encoded video data and the generated group attribute information to generate a packet group and transmitting the packet group to the reception apparatus. Receiving means for receiving a packet group transmitted from the transmission device according to claim 4;
When the group attribute information included in the packet group received by the receiving unit represents a moving image mode in which the motion of the video represented by the picture group included in the group corresponding to the group attribute information is large, the space A video using a first function including a procedure for selecting packets with a weight on a time layer rather than a layer, and the group attribute information is represented by a picture group included in a group corresponding to the group attribute information In the case of a still image mode in which the movement of the image is small, a group corresponding to each of the group attribute information is used by using a second function including a procedure for selecting packets with a weight in the space layer rather than the time layer. Extracting means for extracting a packet to be transmitted from a packet group included in
A relay device comprising: a transfer unit that transfers the packet extracted by the extraction unit to a receiving device that is a destination of the packet.

Images