JP6089970B2 - Video data reduction device and playback device - Google Patents

Description

  The present invention relates to a technique for storing and reproducing video information.

  Like the MPEG (Moving Picture Expert Group) -2 system used in terrestrial digital broadcasting, video signals of digital broadcasting are compressed and transmitted in units of a plurality of images called GOP (Group Of Pictures). In digital broadcasting, individual streams such as video, audio, and other data are multiplexed and transmitted by a multiple signal system called TS (Transport Stream).

  In recent years, devices for storing and reproducing digital broadcasts in a storage device such as a hard disk have become widespread. Since digital broadcasting has a large amount of data, techniques for reducing the amount of data and increasing the number of programs that can be stored are being studied.

  Patent Document 1 discloses a method for reducing the amount of accumulated data by identifying GOPs having the same contents from accumulated video data and deleting repeated GOPs.

  Patent Document 2 discloses a method for reducing the amount of data to be accumulated by deleting streams other than video and audio from the TS.

Japanese translation of PCT publication No. 2006-528862 (FIG. 2A) Japanese Patent Laying-Open No. 2012-186815 (FIG. 1)

  However, only the GOP having the same contents can be deleted by the method described in Patent Document 1. For example, in replay used in sports programs, the replay video and the original video often have the same content. However, the GOP boundaries do not always match between the replay video and the original video, and the GOPs do not have the same contents when the GOP boundaries are different. Therefore, the method of Patent Document 1 can be applied only in a limited case.

  Also, with the method described in Patent Document 2, video stream data with a large amount of data cannot be deleted, and thus the data amount cannot be significantly reduced.

  The present invention has been made to solve the above-described problems, and an object of the present invention is to reduce the amount of video data to be stored even when the GOP boundaries do not match.

  The video data reduction device according to the present invention relates to the content of the video stream of the input stream with respect to a plurality of input streams that are transport streams including a video stream compressed and encoded in units of an image group (GOP) including a plurality of images. The same content among the GOPs related to the same content section with respect to the video identification unit that identifies the section having the same content of the video stream of the other input stream as the same content section and the input stream that identifies the same content section by the video identification unit A data deletion unit that creates a data reduction stream that deletes data of a GOP video stream included in the same content section while leaving data of a GOP video stream that includes only part of the section, and at the time of reproducing the data reduction stream Images included in the GOP of the GOP used for the referenced playback Of the reproduction information generating unit that generates information specifying the image to be reproduced, but with the.

  The playback apparatus according to the present invention includes a transport including a video stream in which a GOP in a section is deleted from a plurality of GOPs included in a video stream compressed and encoded in units of image groups (GOPs) including a plurality of images. A playback device for playing back a stream (TS), which is referred to during playback of the TS, plays back using the GOP of the TS to be played back, and has the same content as the deleted partial section included in the other TS A reproduction information analysis unit that analyzes metadata including information specifying an image to be reproduced using the GOP of the stream, and a TS or other target to be reproduced based on the information that identifies the image analyzed by the reproduction information analysis unit A media playback unit that decodes the TS and generates a video signal.

  According to the present invention, the amount of accumulated video data can be reduced, and the capacity of a storage medium necessary for the accumulation of video data can be reduced.

1 is a configuration diagram of a video data reduction device according to Embodiment 1 of the present invention. FIG. It is an example of distribution of the same content area of each TS of Embodiment 1 of this invention. It is an example of the same content area and identification area information of Embodiment 1 of this invention. It is an example of GOP which can be deleted of Embodiment 1 of this invention. It is an example of the deletion of the video stream of the same content area of Embodiment 1 of this invention. It is an example of the stream Bs of Embodiment 1 of this invention. It is a specific example of the GOP boundary determination process in the metadata creation process of Embodiment 1 of this invention. It is an example of the metadata production | generation process of Embodiment 1 of this invention. BRIEF DESCRIPTION OF THE DRAWINGS It is a block diagram of the reproducing | regenerating apparatus based on Embodiment 1 of this invention. It is a block diagram of the video data reduction apparatus which concerns on Embodiment 2 of this invention. It is a block diagram of the video data reduction apparatus which concerns on Embodiment 3 of this invention. It is an example of the reference stream of Embodiment 3 of this invention.

  Embodiments of the present invention will be described below with reference to the drawings. In the drawings referred to below, the same or corresponding parts are denoted by the same reference numerals. In the following description, MPEG-2 TS will be described as an example of a transport stream (hereinafter, transport stream is also referred to as TS) which is a stream signal obtained by multiplexing a video stream, an audio stream, and the like. The compression method is not limited to MPEG-2. H.264, H.C. Any other video compression method may be used as long as it is a method for compression encoding in units of image groups (GOP) such as H.265. In the following, a case where the number of images (frames) included in the GOP is fixed will be described as an example.

Embodiment 1 FIG.
FIG. 1 is a block diagram of a video data reduction apparatus according to Embodiment 1 of the present invention. The video data reduction device 100 is connected to the data storage device 300. The data storage device 300 is a large-capacity storage device. It is assumed that data stored in the data storage device 300 is stored with a stream ID, which is TS identification information, added thereto. Here, the stream ID is TS identification information determined by a method such as assigning a number in order when recording a TS, or generating from a recording date and time and a channel number in the case of TV broadcast recording. This information makes it possible to identify individual TSs.

  The video extraction unit 101 of the video data reduction device 100 receives two TSs (stream A and stream B in FIG. 1, which are input streams) read from the data storage device 300, and inputs from each TS. The video stream is separated and decoded, and video signal A and video signal B corresponding to stream A and stream B are output. In the video data reduction apparatus 100 of this embodiment, the stream A in FIG. 1 is a reference stream, and the stream B is a reduction target stream. The feature amount extraction unit 102 receives the video signal A and the video signal B output from the video extraction unit 101, extracts the feature amount A and the feature amount B from the video signal A and the video signal B, respectively, and outputs them.

  The video identification unit 103 identifies a section having the same content in each video signal using the feature amount A and the feature amount B input from the feature amount extraction unit 102, and outputs the identified section information as identification section information. The data deletion unit 104 receives the identification interval information output from the video identification unit 103, outputs a stream B request based on the identification interval information, and the acquired stream B is determined to have the same content as the stream A. The data of the video stream is deleted, and a stream Bs that is a TS different from the stream B is created and output. The metadata creation unit (playback information creation unit) 105 outputs a stream A request based on the identification section information output by the video identification unit 103, and from the acquired stream A and the stream Bs output by the data deletion unit, Metadata for reproducing the stream Bs is created and output.

  The control unit 110 is a control unit that controls the operation of the video data reduction device 100, and outputs a data amount reduction instruction to the storage device access unit 111 based on an instruction from the user or a setting. Here, the setting is, for example, a setting for reducing the amount of data for different times of broadcasting of the same program. In this case, the amount of data is reduced by using the TS of the times broadcast later as stream B. The data amount reduction instruction is output.

The storage device access unit 111 reads out a TS having the stream ID specified by the data amount reduction instruction from the data storage device 300 and inputs the TS to the video extraction unit 101. Further, according to the stream B request output from the data deletion unit, the TS of stream B is acquired from the data storage device and input to the data deletion unit 104. Further, according to the stream A request output from the metadata creation unit 105, the TS of the stream A is acquired from the data storage device and input to the metadata creation unit 105. Further, the stream Bs output from the data deletion unit 104 and the metadata output from the metadata creation unit 105 are stored in the data storage device 300.
As described above, in the video data reduction device 100 of this embodiment, the stream A of FIG. 1 is used as the reference stream TS, and the video stream data of the stream B that is the TS of the stream to be reduced is reduced.

Next, the operation of the video data reduction apparatus 100 of this embodiment will be described. In the following description, for convenience, the stream ID of the TS of the reference stream will be described as X, and the stream ID of the TS of the reduction target stream will be described as Y.
The control unit 110 issues a data amount reduction instruction to the storage device access unit 111 so that the TS with stream ID X is stream A and the TS with stream ID Y is stream B is executed. The storage device access unit 111 acquires each TS instructed by the data amount reduction instruction from the data storage device 300 and outputs it to the video extraction unit 101 as stream A and stream B together with the stream ID.

  The video extraction unit 101 includes H.264, H.C. Similarly to a decoder such as H.265, the stream A and the stream B, which are MPEG-2 TS, are respectively decoded, and the video signal A and the video signal B are input to the feature amount extraction unit 102.

The feature amount extraction unit 102 extracts a video signature as a feature amount of each input video signal. Video Signature is an international standard for moving image features defined by ISO / IEC 15938-3 / Amd.4. When the video signature is the same, it can be determined that the video signature has the same content. In the present invention, the feature quantity is not limited to the video signature, and a feature quantity other than the video signature may be used. In this embodiment, the video signature is the feature quantity. The feature amount extraction unit 102 extracts and outputs the feature amount for each of the video signal A and the video signal B.
Note that the processing from the acquisition of stream A from the data storage device 300 to the feature amount extraction by the feature amount extraction unit 102 and the processing from the acquisition of stream B to the feature amount extraction may be performed in order or in parallel. You can go.

The video identification unit 103 receives the feature amount of each video signal output from the feature amount extraction unit 102 and stores it inside. When the input of the feature amount from the feature amount extraction unit 102 is completed, the video identification unit 103 compares and identifies the feature amount of the video signal A and the feature amount of the video signal B, and the contents of the video signal A and the video signal B are identified. The section where is identical is specified. The method described in ISO / IEC 15938-8 / Amd.6 may be used as a method for identifying sections having the same contents.
Note that there may be a plurality of sections having the same content, and the video signal A and the video signal B do not always correspond one-to-one, but may correspond one-to-two or more. For example, as shown in FIG. 2, a and d, i and o-ki, and u and k may be determined as sections having the same contents. Moreover, the length of the section where the content is the same also varies.

The video identification unit 103 outputs information relating to the same content section of the video signal A and the video signal B as identification section information. The identification section information is the number of frames from the first frame of each TS of the start frame and the end frame of the same content sections of the video signal A and the video signal B (hereinafter, the number of frames from the first frame is referred to as a frame number). )including.
For example, as shown in FIG. 3A, the video signal A is composed of twelve frames A0 to A11, and the video signal B is composed of nine frames B0 to B8. When A8 and frames B3 to B6 have the same content, the video identification unit 103 identifies this section as a section having the same content, and creates identification section information as shown in FIG. Here, the frame number of the first frame is 0.

  In FIG. 3, since only one section of the video signal B and one section of the video signal A are the same, information for one section (same content section 1) is shown, but there are a plurality of sections having the same content. In this case, the same information is output as the same content section 2 and the same content section 3 for each section. The identification section information is input to the data deletion unit 104 and the metadata creation unit 105.

  The data deleting unit 104 deletes a part of the video stream of the stream B corresponding to the same content section of the video signal A and the video signal B by using the identification section information in units of GOP. In MPEG-2, compression is performed to reduce the data amount by referring to other frames, but here, reference to frames that exceed the GOP boundary is not performed.

The data deleting unit 104 that has received the identification section information first determines whether the GOP boundary first appears in the reference stream or the reduction target stream for the same content section indicated by the identification section information. In this embodiment, since the number of frames included in the GOP is fixed, the frame number of the start frame of the same content section of each of the reduction target stream and the reference stream included in the identification section information is set as the frame number included in the GOP. It is possible to determine whether the GOP boundary first appears in the stream to be reduced or the reference stream by obtaining the remainder divided by the number of sheets and comparing the size.
When the number of frames included in the GOP is not fixed, information indicating which GOP boundary appears first in the reduction target stream or the reference stream may be added to the identification section information.

  Next, the reduction target stream is acquired from the head, and the reduction target GOP included in the same content section of the reduction target stream is deleted to create a stream Bs that is a data reduction stream. Here, the GOP to be deleted is a GOP that is included in the same content section and in which the GOP boundary appears after the first GOP boundary of the reference stream in the same content section.

  An operation example of the data deletion unit 104 will be described with reference to FIGS. 4, 5, and 6. 4, 5, and 6, only the video stream included in the TS is shown for ease of explanation. Actually, the TS is data in which a plurality of TS packets are multiplexed. Here, for simplicity of explanation, the TS is shown in units of GOP. FIG. 4A illustrates a case where a GOP boundary first appears in the reference stream (stream A) in the same content section, and FIG. 4B illustrates a case where a GOP boundary first appears in the deletion target stream. ing. GOP_A (i) is the i-th GOP from the top of the video stream of stream A (here, the top is 0th), and GOP_B (j) is the j-th GOP from the top of the video stream of stream B Represents.

  FIGS. 5 and 6 are diagrams for explaining the operation of deleting the GOP to be deleted from the stream B, focusing on one section of the same contents of the stream A and the stream B in the case of FIG. It is assumed that the first identical content section 1 when viewed from the beginning of the stream B is between GOP_B (j) and GOP_B (j + 4). The same content section 1 is the same for the frames I to I + K of the stream A, which are the I-th to I + K-th frames from the top of the stream A, and the J- to J + K-th frames J to J + K from the top of the stream B. It is the contents of. Further, the identification section information input to the data deleting unit 104 at this time is as shown in FIG.

  The data deleting unit 104 to which the identification section information shown in FIG. 5 is input first determines whether the GOP boundary first appears in the stream A or the stream B in the same content section 1. Since FIG. 5 shows the case of FIG. 4A, it is determined that the GOP boundary appears first in stream A.

Next, the stream B is acquired from the data storage device 300. A stream B request including the stream ID Y of stream B, frame number 0, and frame number J + K is output to the storage device access unit 111, and the storage device access unit 111 is requested from the beginning of stream B to the GOP including frame J + K. . The storage device access unit 111 acquires the specified range of the stream B from the data storage device 300 and inputs it to the data deletion unit 104.
The storage device access unit 111 specifies the GOP including the frame J + K by reading the stream B stored in the data storage device 300 from the data storage device 300 and analyzing the MPEG-2 data.

  The data deletion unit 104 that acquired the requested data of the stream B creates a stream Bs from which the GOP to be deleted is deleted. Since FIG. 5 is an example of the case of FIG. 4A, GOP (J + 1), GOP (J + 2), and GOP (J + 3) in FIG. 5 are GOPs to be deleted. FIG. 6 shows a stream Bs from which these GOPs to be deleted are deleted.

Here, an example of the deletion operation for the first section with the same content (same content section 1) in the stream B is shown, but for the second and subsequent sections with the same content, the frame number 0 specified in the stream B request is The same processing is performed with the frame number of the first frame of the GOP next to the last GOP read out in the processing of the same content section immediately before (that is, the processing of the same content section 1 when processing the same content section 2). .
In the case of FIG. 4B, the data deletion unit 104 performs the same deletion operation as described above with GOP_B (j + 2) and GOP_B (j + 3) in FIG.

  The stream Bs output from the data deletion unit 104 is input to the metadata creation unit 105 and the storage device access unit 111. The storage device access unit 111 adds the stream ID Y to the input stream Bs and stores it in the data storage device 300. Also, the original TS with stream ID Y as stream B is unnecessary and is deleted. Although stream ID Y is added to stream Bs, a stream ID different from Y may be added. In this case, it is necessary to set the same value as the stream ID of the stream Bs specified by the metadata creation unit 105 described later.

  Next, the operation of the metadata creation unit 105 will be described. The metadata created by the metadata creation unit 105 is data indicating for each GOP which TS data should be used for playback, and includes identification section information, a stream Bs that is a data reduction stream, and a reference stream. Created based on stream A.

  The metadata is data including a stream ID, a GOP head byte position, a PAT head byte position, a PMT head byte position, and the number of playback frames. Here, the stream ID indicates the stream ID of the TS used when reproducing the GOP. The GOP head byte position is a byte position from the head of the TS of the TS packet corresponding to the GOP. PAT (Program Association Table) and PMT (Program Map Table) are program configuration information included in TS necessary for reproduction of MPEG-2 TS, and the PAT start byte position and PMT start byte position are the PAT and PMT TS, respectively. This is the byte position from the beginning of the TS of the packet. The number of reproduced frames indicates the number of frames to be reproduced among the frames included in the GOP.

The metadata is used for playback until the GOP boundary of the reference stream is reached beyond the head position of the same content section, the TS used for the reference stream is switched from the GOP boundary of the reference stream, and Based on the rule that the TS used for the stream Bs is switched again from the beginning of the GOP of the stream Bs including the frame following the last frame of the same content section, it is created for each GOP used for reproduction.
The determination of the GOP boundary is as follows. In the same content section of the identification section information, the frame number of the start frame of the stream to be reduced is J, the frame number of the end frame is J + K, the frame number of the start frame of the reference stream is I, and the end The frame number is I + K and the number of frames included in the GOP is F as follows. K + 1 is the number of frames in the same content section.

A remainder Ri is obtained by dividing the frame number I of the reference stream by the number of frames F. This is the number of frames to be reproduced by the GOP of the data reduction stream after F-Ri exceeds the head position of the same content section.
The number of remaining frames in the same content section is K + 1− (F−Ri), and the quotient Qi obtained by dividing this by F is the number of GOPs of the reference stream included in the rest of the same content section.

The stream Bs is used again from the GOP including the frame J + K + 1 of the stream to be reduced. Of the remaining K + 1− (F−Ri) frames in the same content section, the remainder Rj obtained by dividing J + K + 1 by F uses the GOP of the stream Bs. Therefore, in the Qi + 1th GOP using the reference stream, K + 1− ( F−Ri) − (F × Qi) −Rj frames are reproduced.
An example of processing for determining the GOP boundary will be described with reference to FIG. In FIG. 7, the number of frames included in one GOP is 30, and 93 frames from frame 89 to frame 181 of the reference stream (stream A) and from frame 118 to frame 210 of the stream to be reduced (stream B). Are the same content. That is, F = 30, I = 89, J = 118, K = 92.

For stream A, the remainder of dividing frame number I = 89 by F = 30 is Ri = 29. F-Ri = 30-29 = 1 is calculated, and after exceeding the head position of the same content section 1, one frame is reproduced from the GOP of the stream Bs.
The number of remaining frames in the same content section is K + 1− (F−Ri) = 93−1 = 92 frames, and 92 is divided by 30 and Gi of Qi = 3 stream A is included in the 92 frames. A section with the same content ends in the middle of the fourth GOP.

The stream Bs is used again from the GOP including the frame next to the end frame of the same content section of the stream B. Since J + K + 1 = 211 divided by 30 is Rj = 1, K + 1− (F−Ri) − (F × Qi) −Rj = 93−1−90−1 = 1, and the fourth stream A In GOP, it can be seen that one frame needs to be reproduced.
As described above, since the frame number J + K = 210 obtained by correcting a frame that is actually deleted and does not exist in the stream Bs can be obtained, the same processing can be performed for the subsequent same content section.

An example of metadata creation operation performed by the metadata creation unit 105 will be described with reference to FIG. When the metadata creation unit 105 receives the stream Bs from the data deletion unit 104, the metadata creation unit 105 starts creating metadata from the video identification unit 103 based on the identification section information.
FIG. 8 shows an example of metadata creation corresponding to FIG. In the metadata shown in FIG. 8, MD_A indicates metadata about the GOP of the stream A, MD_B indicates metadata about the GOP of the stream Bs, and parentheses indicate the number of each GOP from the head of each stream. It is shown whether it is a GOP.

  First, the boundary at which the stream Bs and the stream A are switched is determined by the process described with reference to FIG. Since the stream Bs is used for reproduction up to the frame J of the GOP_B (j) of the stream Bs, information included in the MD_B in FIG. 8 is determined based on the stream Bs, and MD_B (0) to MD_B (j) are created. Note that the number of frames up to MD_B (j−1) is the same as the number of frames included in the GOP. The number of frames of MD_B (j) is the number of frames of GOP_B (j) up to frame J.

  Next, in the reproduction of the frame corresponding to the frame J + 1 to the frame J + K of the stream Bs, since the frame A + 1 to the frame I + K−1 of the stream A is used, the metadata creation unit 105 has the same content as the storage device access unit 111. In order to acquire the TS data of the stream A corresponding to the section 1, a stream A request including the stream ID of the stream A is output to the storage device access unit 111. The storage device access unit 111 acquires the TS of the designated stream ID from the data storage device 300 and outputs the TS as the stream A to the metadata creation unit 105.

The metadata creation unit 105 creates metadata MD_A (i + 1), MD_A (i + 2), MD_A (i + 3), and MD_A (i + 4) based on the input stream A. In MD_A (i + 4), there is a possibility that not all frames included in GOP_A (i + 4) are reproduced, but a frame in the middle is reached. Next, the metadata creation unit 105 creates MD_B (j + 4) again based on the stream Bs.
The PAT head byte position and the PMT head byte position can be obtained by analyzing the stream A and the stream Bs.

  The metadata created by the metadata creation unit 105 in this way is output to the storage device access unit 111, and the storage device access unit 111 stores the received metadata with Y being the stream ID of the stream Bs. Save to device 300.

  Note that the information included in the metadata can be information other than that described above, such as indicating the PAT start byte position or the PMT start byte position as a difference from the GOP start byte position. Further, in the metadata regarding GOPs of the same continuous stream such as MD_A (i + 1) to MD_A (i + 4) shown in FIG. 8, the number of frames to be reproduced is indicated by the cumulative number of frames from MD_A (i + 1), and MD_A ( In i + 2) to MD_A (i + 4), it is possible to omit the GOP first byte position.

The example shown here is an example based on MPEG-2 TS, but this video data reduction apparatus is not limited to MPEG-2, for example, when other stream formats such as MMT (MPEG Media Transport) are used. Can be applied similarly.
In the video data reduction device according to this embodiment, the case where the number of frames included in one GOP is fixed has been described as an example. However, when the number of frames included in one GOP is not fixed, A similar operation can be performed by notifying the metadata creation unit 105 of the information on the number of frames of each GOP deleted from the data deletion unit 103.

  Next, a playback device that plays back the stream Bs created by the video data reduction device 100 described above as a playback target stream will be described. FIG. 9 is a configuration diagram of the playback apparatus 200 according to this embodiment. A metadata analysis unit (reproduction information analysis unit) 201 analyzes input metadata and outputs a metadata analysis result. The metadata analysis result output by the metadata analysis unit 201 is input to the media playback unit 202, and the media playback unit 202 outputs an instruction to read TS data (stream data) used for playback based on the metadata analysis result. To do. Then, media signals such as video and audio are reproduced and output from the stream data input in response to the read instruction and the metadata analysis result. The media signal output from the media playback unit 202 is output to the outside of the playback device 200 and input to a display, a speaker, or the like.

  Upon receiving an instruction from the user, the control unit 210 of the playback device 200 outputs a video playback instruction including the stream ID of the TS to be played back to the storage device access unit 211. The storage device access unit 211 of the playback device 200 receives the video playback instruction output from the control unit 210, acquires the metadata corresponding to the specified stream ID from the data storage device 300, and acquires the acquired metadata as metadata. The data is output to the analysis unit 201. In response to a read instruction output from the media playback unit 202, the TS of the designated stream ID is acquired and output to the media playback unit 202 as stream data.

  Next, the operation of the playback apparatus 200 of this embodiment will be described by taking the playback of the TS with the stream ID Y exemplified in the description of the operation of the video data reduction apparatus 100 as an example. The control unit 210 of the playback device 200 outputs a video playback instruction including the stream ID (ie, Y) of the TS to be played back in response to an instruction from the user to the storage device access unit 211. The storage device access unit 211, to which the video playback instruction is input from the control unit 210, reads the metadata (metadata shown in FIG. 8) to which the specified stream ID Y is added from the data storage device 300, and The data is output to the metadata analysis unit 201.

  The metadata analysis unit 201 analyzes the input metadata, and from the data corresponding to each GOP, the TS stream ID, GOP first byte position, PAT first byte position, and PMT first byte used for reproduction of each GOP. The position and the number of GOP playback frames are acquired and output to the media playback unit 201 as a metadata analysis result. The metadata analysis may be sequentially performed according to video reproduction, or the analysis of all metadata may be completed at once.

  Next, the operation of the media playback unit 202 will be described. Upon receiving the metadata analysis result input from the metadata analysis unit 201, the media playback unit 201 reads the PAT and PMT of the TS with the specified stream ID, based on the metadata analysis result, and the stream ID and PAT head. The storage device access unit 211 is requested by indicating the byte position and the PMT first byte position by a read instruction. The storage device access unit 211 acquires the PAT start byte position and the PAT start byte position specified by the PMT start byte position from the TS of the specified stream ID stored in the data storage device 300, and outputs them to the media playback unit 202 To do. Next, the media playback unit 202 stores the GOP data of the TS of the stream ID specified by the metadata analysis result so as to be acquired from the data storage device 300 by indicating the stream ID and the GOP head byte position by a read instruction. Request to the device access unit 211. The storage device access unit 211 acquires data from the designated GOP head byte position of the TS of the requested stream ID, and inputs the data to the media playback unit 202 as stream data.

Based on the PAT, PMT, and GOP data acquired in this way, the media playback unit 202 decodes the number of frames specified by the number of playback frames and outputs it as a media signal including a video signal to be displayed on the display device. To do. When the processing of one GOP is completed, the processing of the next GOP is similarly performed based on the metadata.
The metadata created by the video data reduction device 100 is data for each GOP, and the media playback unit 202 can obtain information such as the stream ID of the TS used in playback of each GOP and the GOP head byte position. Therefore, the playback device 200 can play back the TS whose data has been reduced by the video data reduction device 100.

As described above, the video data reduction device of this embodiment specifies the same content section for two TSs, and reduces video data in units of GOPs included in the same content section from the TS to be reduced. As a result, the amount of video data to be stored can be reduced even when the start and end of the same content section do not coincide with the GOP boundary.
Further, since the metadata for reproducing the TS with the video data reduced is created for each GOP, the TS with the video data reduced can be reproduced using the reproduction apparatus of this embodiment.
In addition, by including information indicating the position in the TS used when reproducing the program configuration information such as PAT and PMT in the metadata, the program configuration information can be easily acquired during the reproduction.

MPEG-2 and H.264 In a compression scheme such as H.264, the stream order may be different from the playback order. In such a case, the playback may be performed according to the playback order instead of the stream order.
In this embodiment described above, for the sake of simplicity, an example has been shown in which two TSs, one reference stream and one deletion stream, are used. However, data reduction using three or more TSs You may make it do.

Embodiment 2. FIG.
FIG. 10 is a block diagram showing a video data reduction apparatus according to Embodiment 2 of the present invention. In FIG. 10, a video extraction unit 101, a feature amount extraction unit 102, and a video identification unit 103 are the same as those in the first embodiment shown in FIG. The stream dividing unit 106 divides a divided stream that is an input TS, and outputs the divided stream as a stream A and a stream B (input stream). Further, the stream dividing unit 106 outputs the position (number of frames from the head in the divided stream) when dividing the divided stream into the stream A and the stream B as division information.

  The data deletion unit 104b receives the stream B output from the stream division unit 106, deletes data based on the identification section information output from the video identification unit 103 for the input stream B, and outputs the data as a stream Bs. . The stream combining unit 107 combines the stream A output from the stream dividing unit 106 and the stream Bs output from the data deleting unit 104b, and outputs the combined stream S.

  The metadata creation unit 105b reproduces the stream S based on the stream S output from the stream combination unit 107, the division information output from the stream division unit 106, and the identification section information output from the video identification unit 103. Create and output metadata. Based on a user instruction or the like, the control unit 110b outputs a data amount reduction instruction including the stream ID of the data reduction target stream to the storage device access unit 111b.

  The storage device access unit 111b reads the TS with the stream ID specified by the data amount reduction instruction from the data storage device 300, outputs the TS as a divided stream to the stream division unit 106, and outputs the stream S and meta The metadata created by the data creation unit 105b is stored in the data storage device 300.

  Next, the operation of the video data reduction apparatus 100 of this embodiment will be described with a focus on differences from the first embodiment. The control unit 110b of the video data reduction device 100b inputs a data amount reduction instruction including the stream ID (referred to as Z) of the data reduction target stream to the storage device access unit 111b. The storage device access unit 111b acquires the TS to which the stream ID specified by the data amount reduction instruction is added from the data storage device 300, and inputs the TS to the stream division unit 106 as a divided stream.

  The stream dividing unit 106 divides the divided stream into two in the first half and the second half. However, the division stream is divided at the GOP boundary. The stream dividing unit 106 outputs the divided first half as stream A and the second half as stream B to the video extracting unit 101. In addition, the number of the last frame of the first half divided into two from the head of the frame (the first frame is set to 0) is output as split information.

  The operations of the video extraction unit 101, the feature amount extraction unit 102, and the video identification unit 103 to which the stream A and the stream B are input are the same as those in the first embodiment. The identification section information output from the video identification unit 103 is input to the data deletion unit 104b and the metadata creation unit 105b.

  The stream B output from the stream segmentation unit 106 is also input to the data deletion unit 104b. The data deletion unit 104b holds the input stream B and is based on the identification section information input from the video identification unit 103. Then, a part of the input video stream of stream B is deleted. The operation of deleting a part of the video stream is the same as the operation of the data deleting unit 104 of the first embodiment. The data deletion unit 104b outputs the TS from which a part of the video stream of the stream B is deleted to the stream combination unit 107 as the stream Bs.

  The stream A output from the stream dividing unit 106 is also input to the stream combining unit 107, and the stream combining unit 107 holds the input stream A. When the stream combining unit 107 receives the stream Bs from the data deleting unit 104b, the stream combining unit 107 combines the stream Bs after the stream A, generates a new TS, and outputs it as a stream S (combined stream). The stream S output from the stream combining unit 107 is input to the metadata creation unit 105b and the storage device access unit 111b. The storage device access unit 111b adds the stream ID Z to the received stream S and stores it in the data storage device 300.

  On the other hand, the metadata creation unit 105b that receives the stream S generates the metadata for reproducing the stream S, the division information received from the stream S and the stream division unit 106, the identification section information received from the video identification unit 103, and the stream S. The process to start is started. The content of the metadata is the same as in the first embodiment. In the first embodiment, the metadata is created from the stream A and the stream Bs. In this embodiment, the metadata is created from the stream S. However, the stream S is a combination of the stream A and the stream Bs. To the first frame in stream S corresponds to stream A, and from which frame corresponds to stream Bs. Therefore, metadata is created by the same processing as in the first embodiment. be able to.

  The metadata created by the metadata creation unit 105b is output to the storage device access unit 111b, and the storage device access unit 111b adds the stream ID of the stream S to the received metadata and stores it in the data storage device 300.

  In this embodiment, since the TS used for reproduction is only the stream S, the stream ID may be omitted from the metadata information.

  The TS in which the data amount is reduced by deleting a part of the video stream generated by the video data reduction device 100b of this embodiment can be played back by a playback device similar to the playback device 200 of the first embodiment. is there.

  As described above, the video data reduction device according to the second embodiment can reduce the TS data amount by identifying a section having the same video content for one TS and deleting a part of the video stream. it can.

  In this embodiment, the divided stream is divided into the first half and the second half, the first half is the stream A and the second half is the stream B. However, the divided stream is divided into three or more, the first one is the stream A, and the rest May be handled as stream B.

Embodiment 3 FIG.
In the first embodiment described above, a section having the same content is specified for two TSs, and one of the sections having the same content is deleted to reduce video data. In Embodiment 2, video data is reduced by specifying a section having the same content in a single TS and deleting one of the sections having the same content. In this embodiment, a reference stream is created from a large number of TSs that are stream to be reduced (input stream), and the same content section as the reference stream is deleted from the TS, thereby reducing the amount of video data. In the following, in order to simplify the description, a case where the number of streams to be reduced is 3 will be described as an example. However, the following description does not limit the number of TSs of the present invention to 3.

  FIG. 11 is a configuration diagram of the video data reduction device 100c of this embodiment. The video extraction unit 101c, the feature amount extraction unit 102c, and the video identification unit 103c are respectively the video extraction unit according to the first embodiment for three streams 1 to 3, video signal 1 to video signal 3, and feature amount 1 to feature amount 3. 101, the feature amount extraction unit 102, and the video identification unit 103 are processed, and the video identification unit 103c outputs identification section information. Streams 1 to 3 are reduction target streams. The identification section information output from the video identification unit 103 c is input to the reference stream creation unit 108. Streams 1 to 3 input to the video extraction unit 101c are input to the reference stream creation unit 108. The reference stream creation unit 108 refers to the reference stream and the reference stream based on the stream 1 to stream 3 and the identification section information. Create and output information.

  The data deletion unit 104c receives the same stream 1 to stream 3 as the video extraction unit 101c and the reference information output from the reference stream creation unit 108, and deletes a part of the video stream for each of the streams 1 to 3 Stream 1S to stream 3S with reduced data amount are output. The metadata creation unit 105c receives the reference stream ID, the reference stream created by the reference stream creation unit 108, and the streams 1S to 3S output from the reference information data deletion unit 104c, and reproduces the streams 1S to 3S, respectively. Metadata 1 to metadata 3 are output.

  The control unit 110c outputs a data amount reduction instruction including stream IDs of a plurality of (three in this embodiment) TSs for reducing video data, and a stream ID of a reference stream created by the reference stream creation unit. The storage device access unit 111c acquires the TS to which the stream ID specified by the data amount reduction instruction input from the control unit is attached from the data storage device 300, and streams the streams 1S to 3S output by the data deletion unit, respectively. 1 to 3 stream IDs are added and stored in the data storage device 300, and the stream IDs of stream 1 to stream 3 are added to the metadata 1 to metadata 3 output by the metadata creation unit 105, respectively. The reference stream ID output from the control unit 110c is added to the reference stream output from the reference stream creation unit and stored in the data storage device 300.

  Next, the operation of the video data reduction device 100c of this embodiment will be described focusing on the differences from the first embodiment. Based on a user instruction or the like, the control unit 110c outputs a data amount reduction instruction including a stream ID of a TS for which video data is to be reduced to the storage device access unit 111c. Also, the reference stream ID that is the stream ID of the reference stream is output to the storage device access unit 111c and the metadata creation unit 105c.

  Receiving the data amount reduction instruction from the control unit 110c, the storage device access unit 111c receives the stream ID of the TS specified in the data amount reduction instruction (here, the stream ID of the TS to be stream 1 is L, and the TS of the TS to be stream 1) TS with stream ID M and stream 3 of stream 3 is acquired from data storage device 300, and stream 1 to stream 3 are extracted from video extraction unit 101c, reference stream creation unit 108, The data is input to the data deletion unit 105c.

  The video extraction unit 101c that has received the stream 3 from the stream 1 extracts the video signals 1 to 3 by the same processing as the video extraction unit 101 of the first embodiment, and inputs them to the feature amount extraction unit 102c. The feature quantity extraction unit 102c inputs the feature quantity 1 to the feature quantity 3 for each video signal to the video identification unit 103c by the same process as the feature quantity extraction unit 102 of the first embodiment. The video identification unit 103c identifies the same content section of the streams 1 to 3 and refers to the identification section information by the same processing as the video identification unit 103 of the first embodiment based on the input feature quantity 1 to feature quantity 3. The data is output to the stream creation unit 108.

  The reference stream creation unit 108 that has received the identification section information creates and outputs a reference stream based on the streams 1 to 3 input from the storage device access unit 111c and the identification section information. FIG. 12 is a diagram illustrating an example of a reference stream created based on stream 1 to stream 3. The video stream of the section having the same contents of stream 1 to stream 3 indicated in the identification section information is extracted from one of stream 1 to stream 3 to create a reference stream that is another TS. Note that when extracting a video stream, GOP is used as a unit so that the entire section having the same content is included. In the reference stream shown in FIG. 12, the blank sections shown before and after each of the sections a, i, c, and c indicate that, when extracted in units of GOP, sections other than the same contents may be included. Yes.

  Also, as shown in FIG. 12, PAT and PMT are extracted from the reduction target stream that is the source when creating the reference stream and included in the reference stream. Further, the reference information includes information on which PAT and PMT each GOP of the reference stream corresponds to. Note that a plurality of PATs and PMTs may exist in one TS, and in playback, it is necessary to refer to the PAT and PMT immediately before the GOP to be played back. The PAT1.1, PMT1.1, PAT1.2, PMT1.2, PAT2.2, and PMT2.2 of the reference stream shown in FIG. 12 are the PAT immediately before the GOP extracted from the original stream to be reduced based on this rule. , PMT is extracted.

  FIG. 12 shows an example in which a video stream is extracted from a young TS and a reference stream is created (extracting a section from stream 1 to a-i-woo and a section from stream 2), but other criteria The location of the same content section used for creating the reference stream may be selected. For example, when the portion having the smallest data amount of the section to be extracted when extracting with GOP as a unit is selected, the data amount of the reference stream can be reduced. Alternatively, with an emphasis on image quality, the average of the GOP code amount (GOP average code amount) may be calculated for each of the same content sections, and a portion with a large GOP average code amount may be selected.

  Further, the reference stream creation unit 108 creates and outputs reference information indicating which part of each of the stream 1 to stream 3 is not the same as which part of the created reference stream. For example, in FIG. 12, the reference information includes information on the start frame and end frame from the beginning of stream 1 in the section of stream 1 and information on the start frame and end frame of the section of the reference stream corresponding to section a. including.

  The reference information output by the reference stream creation unit 108 is input to the data deletion unit 104c. The reference information includes information for identifying the section having the same content as the reference stream for each of the streams 1 to 3, and the identification section information input to the data deletion unit 104 and the metadata creation unit 105 in the first embodiment It has similar information. Based on the reference information input from the reference stream creation unit 108, the data deletion unit 104c performs the same processing as that of the data deletion unit 104 of the first embodiment from each stream 1 to stream 3 with the same content as the reference stream Are deleted, and streams 1S to 3S (data reduced streams) are created and output.

  The reference stream and reference information created by the reference stream creation unit 108 and the streams 1 to 3 created by the data deletion unit are input to the metadata creation unit 105c. Further, the metadata creation unit 105c receives the reference stream ID output from the control unit 110c. The metadata creation unit 105 c generates metadata 1 to metadata 3 based on the reference stream, the reference information, and the streams 1 to 3. Here, the reference stream corresponds to the stream A in the first embodiment, each of the metadata 1 to metadata 3 corresponds to the stream B, the reference information corresponds to the identification section information, and each of the streams 1 to 3 corresponds to the embodiment. The metadata 1 to metadata 3 are created by the same processing as that of the first metadata creation unit 105. Note that the value of the reference stream ID is used as the metadata stream ID.

  The reference stream output from the reference stream creation unit 108, the streams 1S to 3S created by the data deletion unit, and the metadata 1 to metadata 3 created by the metadata creation unit are input to the storage device access unit 111c. . When receiving the reference stream, the storage device access unit 111c adds a reference stream ID and stores the reference stream ID in the storage device 300. When stream 1S to stream 3S are received, the stream IDs of stream 1 to stream 3 are added to each of them and stored in the storage device 300. When the metadata 1 to metadata 3 are received, the stream IDs of the streams 1 to 3 are added and stored in the storage device 300. Note that when storing the streams 1S to 3S in the storage device 300, the streams 1 to 3 are deleted from the storage device 300.

  The TS in which the data amount is reduced by deleting a part of the video stream generated by the video data reduction device 100c of this embodiment can be played back by a playback device similar to the playback device 200 of the first embodiment. is there.

  As described above, according to the video data reduction device of this embodiment, a reference stream is created from a number of TSs, and a section having the same content as the reference stream is deleted from a number of original TSs. The amount of data can be reduced.

  In the above-described third embodiment, the same content section of video is identified from a large number of streams, a reference stream is created, and a part of the original large number of streams is deleted to reduce the TS data amount. When a new program is recorded in a state where a reference stream that has already been created exists, the reference stream can be used as the stream A shown in the first embodiment.

  In the above-described third embodiment, the same content section of the video is specified from a large number of TSs and the reference stream is created. However, when creating the reference stream, for example, without identifying the video for the entire video, The reference stream may be about 3 minutes from the start. As a result, the reference stream includes the opening portion of the program. When the next portion of the same program is recorded, the opening portion can be deleted, and the amount of data can be effectively reduced with a small amount of processing.

  In Embodiment 3 described above, the same content section of the video is specified from a number of streams and the reference stream is created. However, when creating the reference stream, video identification need not necessarily be performed. For example, when a new program starts, it can be considered that the reference stream is about 3 minutes from the start. In this way, since the opening part of the program is included in the reference stream, the opening part can be deleted when the next portion of the same program is recorded, and the amount of data can be effectively reduced with less processing. .

  The video data reduction apparatus according to the first embodiment shown in FIG. 1 can be realized by hardware such as an LSI, or can be realized by software operating on a processor. The data storage device can be configured as a device for storing video data using a hard disk device or the like. In addition, it is possible to configure an apparatus for storing and reproducing video data in combination with a reproduction apparatus. The same applies to the video data reduction apparatus shown in the second and third embodiments.

100, 100b, 100c Video data reduction device, 101, 101c Video extraction unit, 102, 102c Feature amount extraction unit, 103, 103c Video identification unit, 104, 104b, 104c Data deletion unit, 105, 105b, 105c Metadata creation unit , 106 stream division unit, 107 stream combination unit, 108 reference stream creation unit, 110, 110b, 110c control unit, 111, 111b, 111c storage device access unit, 200 playback device, 201 metadata analysis unit, 202 media playback unit, 210 control unit, 211 storage device access unit, 300 data storage device

Claims (8)

For a plurality of input streams that are transport streams including a video stream compressed and encoded in units of an image group including a plurality of images (hereinafter referred to as GOP), the contents of the video stream of the input stream and the other A video identification unit that identifies a section having the same content of the video stream of the input stream as the same content section;
For the input stream in which the same content section is specified by the video identification unit, among the GOPs related to the same content section, leave the video stream data of the GOP that includes only a part of the same content section, A data deletion unit that creates a data reduction stream in which the video stream data of the GOP included in the same content section is deleted;
A reproduction information creation unit that creates information for specifying an image to be reproduced among images included in the GOP of the GOP used for reproduction referred to when reproducing the data reduction stream;
A video data reduction device comprising: The plurality of input streams are one or more reference streams and one or more reduction target streams,
A video extraction unit for decoding the video stream of each of the reference stream and the reduction target stream and outputting a video signal;
A feature amount extraction unit that extracts a feature amount of each image of the video signal output by the video extraction unit;
The video identification unit compares the feature amount extracted by the feature amount extraction unit, and the content of the image of the video signal of the reference stream is the same as the content of the image of the video signal of the reduction target stream. Identify the section as the same content section,
The data deletion unit creates the data reduction stream for the reduction target stream,
The reproduction information creation unit creates metadata including information specifying the image to be reproduced of the GOP of the reference stream and the data reduction stream used for reproduction.
The video data reduction device according to claim 1.   3. The video data reduction device according to claim 2, wherein when the transport stream includes program configuration information, the metadata includes position information of the program configuration information in the data reduction stream and the reference stream. . A stream splitting unit that splits one transport stream into two or more to generate the input stream;
At least one of the input streams is set as a reference stream, and at least one of the input streams different from the reference stream is set as a reduction target stream, and the video streams of the reference stream and the reduction target stream are decoded. A video extraction unit that outputs a video signal,
A feature amount extraction unit that extracts a feature amount of each image of the video signal output by the video extraction unit;
A stream combining unit that combines the reference stream and the data reduction stream to output a combined stream;
The video identification unit compares the feature amount extracted by the feature amount extraction unit, and the content of the image of the video signal of the reference stream is the same as the content of the image of the video signal of the reduction target stream. Identify the section as the same content section,
The data deletion unit creates the data reduction stream for the reduction target stream,
The reproduction information creating unit creates metadata including information specifying the image to be reproduced that is referred to when the combined stream is reproduced;
The video data reduction device according to claim 1. The plurality of input streams are streams to be reduced,
A video extraction unit that decodes the video stream of each of the plurality of reduction target streams and outputs a video signal;
A feature amount extraction unit that extracts a feature amount of each image of the video signal output by the video extraction unit;
A reference stream is created by extracting data of the video stream of the GOP that includes a part or the whole of the same content section specified by the video identification unit from any of the plurality of reduction target streams. And a reference stream creation unit that creates reference information indicating a correspondence relationship between the same content sections of the plurality of reduction target streams,
The video identification unit compares the feature amounts output from the feature amount extraction unit, and for the plurality of reduction target streams, the image content of the video signal of the reduction target stream and other reduction target streams. Identifying a section in which the content of the image of the video signal is the same as the same content section,
The data deletion unit creates the data reduction stream for each of the plurality of reduction target streams based on the reference information,
The reproduction information creation unit creates metadata including information specifying the image to be reproduced that is referred to when each of the plurality of data reduction streams is reproduced;
The video data reduction device according to claim 1.   The reference stream creating unit compares the data amount of the same content section, selects the same content section of the stream to be reduced whose data amount is smaller than the other, and creates the reference stream. The video data reduction device according to claim 5.   The reference stream creation unit compares the average code amount of the GOP in the same content section, selects the same content section of the stream to be reduced whose average code amount is larger than the other, and the reference stream The video data reduction device according to claim 5, wherein the video data reduction device is created. A transport including the video stream from which a part of the GOP is deleted from among the plurality of GOPs included in the video stream compressed and encoded in units of an image group including a plurality of images (hereinafter referred to as GOP). A playback device for playing back a stream (hereinafter referred to as TS),
The GOP of the video stream having the same content as the deleted partial section included in the other TS and the image that is referred to when the TS is played and is played back using the GOP of the TS to be played back A reproduction information analysis unit that analyzes metadata including information specifying the image to be reproduced;
A media playback unit that generates a video signal by decoding the TS to be played back or the other TS, based on information specifying the image analyzed by the playback information analysis unit;
A playback apparatus comprising:

Images