JP2008092426A - Apparatus and method for reproducing video information recorded in video information recording medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a video information reproducing apparatus which achieves smooth reverse reproduction even in MPEG4-AVC. <P>SOLUTION: A picture to be an access point specified as a random accessible position is set as an I picture or a P picture. Before starting reverse reproduction, the picture of the access point is stored in a frame memory prepared for reverse reproduction which differs from a frame memory of a decoder, and in the reverse reproduction, pictures after the access point are decoded while referring to the data stored in the frame memory. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a randomly accessible video information recording medium, a recording apparatus and a recording method for recording video data on a video information recording medium, and a reproducing apparatus and a reproducing method for reproducing video data from a randomly accessible video information recording medium. It is about.

When playing back encoded image data written in a storage medium in a format compliant with the MPEG system, a method for smooth playback in the reverse direction is the method after decoding all frames included in the GOP. A method realized by storing picture data has been proposed (for example, Patent Document 1).
JP-A-8-32935

  In recent years, new encoding schemes, such as MPEG4-AVC (H.264), that can ensure sufficient image quality even when encoded at a low bit rate have begun to spread. In order to realize high image quality at a low bit rate, it is necessary to reduce the number of I pictures having a large code amount as much as possible. On the other hand, since the first frame of a GOP must be an I picture, reducing the number of I pictures is equivalent to increasing the GOP interval. For example, in “1-segment broadcasting”, which is a television broadcast for portable terminals, a GOP of up to 5 seconds is allowed. In such a long-time GOP, it is an unrealistic method in terms of cost to hold all decoded frame data in the GOP in the memory as in the above-mentioned document.

Further, as a general MPEG restriction condition, it is not possible to reproduce from a picture in the middle of a GOP. It is necessary to reproduce from the first I picture. In order to reproduce a picture in the middle of a GOP, it is necessary to decode at least all I pictures and P pictures from the I picture at the head of the GOP to the picture to be reproduced. For example, some DVD players perform reverse playback by repeatedly decoding an I picture and a P picture at high speed and decoding a target picture. In DVD, since the GOP length is set to be relatively short, reverse playback using such a method is possible. However, if the GOP becomes longer as described above, the number of P pictures up to the target picture becomes smaller. Therefore, a decoder capable of decoding at a speed that is ten times faster than usual is required, which is not practical.

In the case of MPEG4-AVC, the decoding process is more complicated than that of MPEG2, and therefore, the time required for decoding is several times that of MPEG2. Furthermore, since the picture reference rules at the time of encoding are very complicated, the target frame cannot be decoded by simply decoding the I picture and P picture. Therefore, smooth reverse reproduction of the MPEG4-AVC stream is practically impossible, and there is a problem that the user cannot perform reverse reproduction similar to MPEG2.

  The present invention has been made to solve the above-described problems, and provides a video information recording medium, a playback apparatus, and a playback method that enable smooth reverse playback even in MPEG4-AVC streams. Objective.

  The present invention predicts an I picture that is an intra-frame coded image, a P picture that is a prediction coded image of a set of blocks predicted from one frame, and a block set that is predicted from two frames. A video information recording medium on which video data composed of video units consisting of B pictures, which are encoded images, is recorded. Pictures subsequent to a P picture designated as an access point are P pictures designated as access points. Without referring to the previous picture, it has information describing whether the P picture in the GOP is a P picture as an access point, a P picture necessary for decoding the access point, or not required, Decode to save the P picture that will be the decoded access point The leader is a video information reproducing apparatus comprising: a frame memory prepared separately.

  According to the present invention, smooth reverse reproduction can be realized even in a stream encoded by MPEG4-AVC.

Hereinafter, embodiments will be described with reference to the drawings.
Although the case where it is an optical disk as a video information recording medium is mainly shown, it may be a recording medium such as a hard disk or a semiconductor memory. Also, in the present embodiment, a case will be described in which recording is performed by compressing digital video data using an MPEG4-AVC (Advanced Video Codec) encoding method. Each frame of the digital video data is composed of any one of three types of encoded images of I picture, P picture, and B picture. An I picture is an intra-frame encoded image that is encoded within one frame. A P picture is a predictive coded image in which one frame is divided into a plurality of blocks and is predicted from another frame for each block, that is, an aggregate of blocks predicted from one frame. It is. A B picture is a predictive coded image in which one frame is divided into a plurality of blocks and prediction is performed with reference to another two frames for each block, that is, a block predicted from two frames. It is an aggregate. Further, a GOP (Group of Picture) is configured as a video unit including at least one I picture and including one or more P pictures and one or more B pictures starting from the I picture, The video data is composed of a plurality of GOPs. However, although GOP is not defined by the standard in MPEG4-AVC, the GOP concept applied to MPEG4-AVC is referred to as GOP here and will be described.

Embodiment 1 FIG.
FIG. 1a is a diagram for explaining the configuration of a GOP. In the figure, I represents an I picture, P represents a P picture, and B represents a B picture. The leftmost I1 picture is a picture with the oldest time in time, and the picture with the newest time goes to the right and is arranged in the display order. In the figure, arrows indicate the relationship between prediction and reference when encoding each picture. The numbers following I, P, and B are numbers for identifying each picture, and the numbers are assigned to I, P, and B from the oldest in time.

  FIG. 1a is a configuration example of a GOP when encoded according to MPEG4-AVC. A procedure for decoding each picture will be described with reference to FIG. Although the display order is as shown in FIG. 1, when recording on a recording medium or the like, they are recorded in the decoding order. For example, in FIG. 1a, recording is performed on the recording medium in the order of I1, P1, B1, B2, B3, B4, P2,.

  In FIG. 1a, I1 is a picture that can be decoded independently and is the first picture, so it is first decoded. Since the decoded I picture is necessary for subsequent decoding, it is temporarily stored in the frame memory in the decoder. Next, P1 is decoded, and the decoded P picture is also temporarily stored in the frame memory. Subsequently, B1 is decoded with reference to the already decoded I1 and P1, and temporarily stored in the frame memory. Here, I1 in the frame memory is discarded. Next, referring to B1 and P1, B2 is decoded and temporarily stored in the frame memory. Then, B3 is decoded with reference to P1 and B2, temporarily stored in the frame memory, and B3 in the frame memory is discarded. Next, referring to P1 and B3, B4 is decoded and temporarily stored in the frame memory, and the picture in the frame memory is discarded except for B4. Next, P2 predicted from B4 is decoded and temporarily stored in the frame memory. In the same manner, decoding progresses one after another.

  As described above, decoding one picture requires temporally forward and / or backward pictures, but this mechanism is a major obstacle to reverse playback. Next, the case of reverse playback from P2 will be described. B4 is required for decoding P2, and B3 and P1 are required for decoding B4. P1 and B2 are required for decoding B3, I1 is required for decoding P1, and B1 and P1 are required for decoding B2. That is, in order to decode P2, it is necessary to decode all the pictures from I1 to B4, and all the processes described above are executed. The next display of B4 has already been decoded and is displayed using the image stored in the frame memory in the decoder, but since the data of B3 does not remain in the frame memory, it is decoded again from I1. I need to fix it. In the same manner, reverse playback is performed one by one. However, there are many cases where the video stored in the frame buffer in the decoder cannot be used, and it is necessary to re-decode from the I picture at the head of the GOP each time. For this reason, it takes 33 msec or more to decode one frame, resulting in a poor and awkward reverse reproduction.

FIG. 1b differs from FIG. 1a in that P2 refers to P1, but in this case, in order to decode P2, it is only necessary that I1 and P1 can be decoded. However, in the case of MPEG4-AVC, there is no way to know which picture is referring to which picture until decoding is actually started. Therefore, even if P2 makes a reference as shown in FIG. As with FIG. 1a, all pictures need to be decoded. As described above, in the case of MPEG4-AVC, since the number of pictures that need to be decoded when performing reverse reproduction is much larger than that of MPEG2, there is a problem that smooth reverse reproduction is practically impossible.

In order to solve this problem, it is necessary to give a certain restriction at the time of encoding. However, giving a large restriction impairs the goodness of MPEG4-AVC, and should be minimized. FIG. 2 shows a configuration of a GOP that is encoded by adding a certain constraint condition when encoding a picture to be an access point AP. Each picture is arranged in the display order from the left. In this GOP, the following three restrictions are added as conditions for encoding pictures after the access point AP.
1. The picture serving as the access point is an I or P picture.
2. The P picture serving as the access point is predicted from the I picture at the head of the GOP or the P picture before the access point, and there is no B picture among the pictures necessary for decoding the P picture serving as the access point.
3. Pictures after the P picture of the access point are not predicted or referenced from pictures before the access point. However, the I picture at the head of the GOP or the P picture serving as an access point is excluded.

  Here, the access point AP is an accessible position (point) when performing so-called random access for reproduction from an arbitrary position of video information, such as a point in time desired by the user. In other words, if the access point can be decoded, it means that the picture after the access point can be decoded even if there is no video data before the access point. Under the above conditions, for example, it is prohibited for P2 to refer to B5 or for B12 to refer to P3. In the case of a read-only optical disc, the position is specified at the time of authoring the disc, and in the case of a recordable or rewritable optical disc, the recording device automatically designates when recording video data on the optical disc.

  In FIG. 2, when encoding is performed under the above-described conditions, in order to perform reproduction from P4 which is the access point AP, first, I1 at the beginning of the GOP is decoded, and only P picture of P1 and P2 is decoded. Decoding is possible. Since the pictures after P4 are not used for referring to and predicting pictures before P4, subsequent pictures can be reproduced continuously by decoding with a normal sequence. When encoding is performed without any restrictions, it is necessary to decode 14 sheets before P4 as already described. However, by providing a P picture as an access point, the number of pictures to be decoded can be greatly reduced.

  Under the above encoding conditions, the following index information is arranged in the video data in order to perform quick reproduction from the middle of the stream. FIG. 3 is a structure diagram of index information for accessing the head of the GOP in the video data. FIG. 4 shows the index information syntax. In FIG. 3, Entry_map () is a data area in which information necessary for accessing the I picture at the head of the GOP is stored, and constitutes a part of navigation data. Navigation data refers to general control information and management information for controlling the reproduction of content such as video files in a recording medium, and includes index information for I pictures.

  In “number_of_IAP” in FIG. 4, the total number of I picture access points existing in the moving image file is described. Although the video file is composed of a plurality of GOPs, the number of GOPs and the number of “number_of_IAP” do not need to match because the first I picture of the GOP does not necessarily have to be an access point, but the total number of GOPs Never exceed. In the description of this embodiment, the description will be made on the assumption that the total number of GOPs and the total number of “number_of_IAP” are the same.

In FIG. 4, the for loop statement next to “number_of_IAP” is a loop that is repeated by the number of “number_of_IAP”. In this, access point information from the access point of one I picture to the access point of the next I picture is described. In the case of a GOP of about several seconds, since the I picture at the head of the GOP is always designated as an access point, it may be regarded as access point information for each GOP.
“I_PTS_AP” in FIG. 4 describes the Presentation Time of the I picture, which is display time information indicating the playback timing at the head of the access point. The Presentation Time here may be a PTS described in each MPEG2 picture, or may be a value indicating a relative time from the beginning of the video file. The next “I_SCN_AP” is information indicating the position information in the video file or the position information in the disc for the I picture at the head of the access point. In the present embodiment, the number of relative sectors from the head address of the video file for the I picture at the head of the access point is described. Here, the number of sectors is used, but the number of bytes may be used as long as it is information that can specify the position from the top of the video file or the absolute position of the disk. “Size_of_IAP” is information indicating the data size of the I picture at the head of the access point. Here, the relative number of sectors from the beginning of the GOP in which the I picture exists is described for the sector in which the last byte of the I picture exists. With the above three pieces of information, it is possible to know the head position, display time, and data size of the I picture at the head of the access point.

  By configuring the index information of the I picture access point as described above, the position information, time information, and picture size of the access point of each I picture can be known.

Next, a method of selecting a P picture necessary for decoding the access point P5 from a series of picture data continuously read will be described with reference to FIG. FIG. 5 is a diagram showing the structure of “GOP_structure” in which picture information in each GOP is described. “Number_of_P_pictures” describes the total number of P pictures existing between access points. The next for loop is a loop that is repeated by the number of “number_of_P_pictures”. “Picture_type” describes attribute information indicating whether each P picture is necessary for decoding the P picture of the next access point. For example, “01” is described when the picture is necessary, “00” when it is not necessary, and “11” is described when the P picture is an access point, “P_SCN” is from the GOP head address to the head of the P picture. Since the selection of the picture is automatically processed in the LSI, “GOP_structure” is placed at the head of the GOP in the moving image file.In the case of MPEG4-AVC, SEI (Supplemental Enhancement In the description of the present embodiment, a case where it is arranged in the SEI will be described, but it is not always necessary to exist in the stream. The same effect can be obtained even if it is placed in Entry_map () ".

  As described above, by adding the attribute information of the P picture in the GOP, it is possible to know in advance the minimum necessary P picture before decoding the P picture serving as an access point before starting the GOP decoding. It becomes possible.

Next, a procedure for performing reverse reproduction using the index information and information of the P picture in the GOP will be described. FIG. 6 is a diagram showing the configuration of the playback device 600. When a reverse playback instruction from the user is input to the CPU 602 from the user interface (I / F) unit 601, a navigation data 1 read command is issued to the drive control unit 604 that controls the drive 603 that reads data from the optical disk. . The read navigation data 1 is transferred to the work memory 605 via the drive control unit 604. The CPU 602, the drive control unit 604, and the work memory 605 are connected to the CPU 602 by a system bus 606 composed of an address bus and a data bus, and commands from the CPU 602 and data transfer between the blocks are transferred through the system bus 106. Done with. The CPU 602 extracts management information related to a program designated for reproduction by the user from the navigation data 1 developed in the work memory 605. In accordance with the extracted management information, the CPU 602 instructs the drive control unit 604 to read data of the video file 3 necessary for playback from the video data 2, and desired data is read by the drive 603. The read data is temporarily stored in the buffer memory 608 via the system bus 606. The CPU 602 performs control so that the buffer memory 608 does not become empty or overflow so that video and audio can be reproduced without interruption. Data once stored in the buffer memory 608 is decoded into video information by a decoder 607 that decodes the encoded data, and is output to a display device 610 such as a TV.

  Next, an operation when reverse playback is designated by the user will be described. In the present embodiment, it is assumed that the reverse playback command is issued from the user in the third frame from the P picture as the second access point in FIG. Further, it is assumed that the bit rate of the stream to be reproduced is 10 Mbps and the reading speed from the disk of the drive is 50 Mbps.

When a user reverse playback command is issued from the user I / F unit 601, the CPU 602 refers to Entry_map () of the navigation data 1 stored in the work memory 605. Here, I_PTS_AP (k) is time information on the I picture that is closest to the currently playing time and is earlier than the currently playing time. The head address of the video file 3 currently being reproduced can be recognized from the file system of the recording medium. Therefore, an address obtained by adding “I_SCN_AP (k)” that is position information about the I picture to the head address of the video file 3 becomes the absolute address of the I picture located at the head of the GOP including the target access point. The CPU 602 instructs the drive control unit 604 to read data from the absolute address. As for the amount of data read out here, it is desirable to read out data for 1 GOP and store it in the buffer memory 608 in order to realize smooth reverse reproduction. This embodiment will be described on the assumption that the buffer memory size is sufficiently larger than the bit rate and that all GOP data is stored in the buffer memory 608. On the other hand, the end address to be read can be calculated as “I_SCN_AP (k + 1) −1” because “I_SCN_AP (k + 1)” is the head address of the next I picture.

  When the reading of GOP data is completed in this way, the CPU 602 reads “GOP_structure” data in the SEI message from the GOP data stored in the buffer memory 608. Since the access point closest to the time when the reverse playback instruction is given is P6, a P picture necessary for decoding P6 is found from “GOP_structure”. In this case, only the I picture is necessary for decoding the first access point P3, and the pictures necessary for decoding P6 are P3 and P4. Therefore, it can be seen that P6 can be decoded by decoding three frames I, P3, and P4. Next, the CPU 602 needs to take out the data of four pictures including the P6 from the buffer memory 608. Generally, since one frame of picture is assigned to one PES packet, the CPU 602 determines the head of the PES packet. By retrieving the start code from the buffer memory 608, the encoded data of each picture can be easily extracted.

  As described above, the data of the four pictures I, P3, P4, and P6 are found from the buffer memory 608 and transferred to the decoder 607. The decoder 607 starts decoding simultaneously with the transfer of these data. The I, P3, and P4 pictures that are decoded access points at this time are stored in the reverse playback frame memory 609 separately from the frame memory necessary for normal decoding until the reverse playback of the GOP is completed. Is done. Furthermore, the CPU 602 transfers the data to the data decoder 607 up to P7, B14, and B13. Since the B pictures of B13 and B14 can be decoded only after P6 and P7 are decoded, the decoded P6 and P7 are stored in the frame memory in the decoder 607. However, the decoded P6 is also stored in the reverse playback frame memory 607 at the same time. Since B13 is not displayed unless B14 is decoded, the decoder 607 stores B14 in the frame memory in the decoder after decoding B14, and then starts decoding of B13. When the decoding of the B14 picture is completed, the decoded video is output from the decoder. At this stage, the image displayed by returning one frame is displayed to the user. Next, when the decoding of B13 is completed, the decoded video of B13 is output from the decoder 607. Next, the video of P6 that has already been decoded is output from the decoder 607. In this way, reverse playback from B14 to P6 is performed. When the display of P6 is completed, the video stored in the frame memory in the decoder 607 is once cleared.

  When the display of P6 is completed, the CPU 102 transfers P4 and P5 to the decoder. The decoder 607 refers to P3 stored in the reverse playback frame memory 609 and decodes P4 and P5. The decoded P4 and P5 are stored in a frame memory in the decoder. After the decoding of P5 is completed, the CPU 602 transfers the data of B11 and B12 to the decoder 607. The decoder 607 decodes B12 with reference to P6 stored in the reverse playback frame memory 609 and P5 stored in the frame memory in the decoder 607. When the decoding of B12 is completed, the video of B12 is output from the decoder 607. After completing the decoding of B12, B11 is decoded and the video of B11 is output. Next, the video of P5 stored in the frame memory in the decoder 607 is output. Similarly, reverse playback video is sequentially output with reference to P5, P4, and P3 already stored in the reverse playback frame memory 609 and the frame memory 607 in the decoder.

  Next, the time until video is output when reverse playback is performed according to the above procedure will be described. First, the time until the B14 frame is displayed will be described. Since the speed of reading data from the disk of the drive 603 is 50 Mbps and the encoding rate of the stream is 10 Mbps, assuming that 1 GOP is 1 second, the time required for reading data for 1 GOP is 200 msec. The time required to transfer I, P3, P4, P6, and P7 to the decoder 607 is small enough to ignore the recent memory bus bandwidth, so the time required for data transfer is not considered. The time required to decode each frame greatly depends on the encoding method. However, assuming that 10 msec per frame is assumed for the sake of simplicity of calculation, the above-described decoding of 5 frames requires 50 msec. Since decoding of B14 requires 10 msec, reverse reproduction starts 200 + 50 + 10 = 260 msec after the user requests reverse reproduction. Decoding of B13 requires 10 msec, but since there is a display time of about 33 msec per frame, B13 can be displayed without delay. Since the decoded data is already stored in the reverse playback frame memory 109, P6 can be displayed 33 msec after displaying B13. Next, since it is necessary to decode P4 and P5 in order to display B12, this takes about 20 msec. Furthermore, since it takes 10 msec to decode B12, B12 can be displayed 30 msec after displaying P6. Since B11 refers to P5 and B12 already stored in the frame memory in the decoder 107, it can be decoded in 10 msec.

Since subsequent frames can be decoded within 33 msec by decoding in the same procedure, the user can watch reverse playback in the same manner as normal playback within the range of playing back GOP once read to the buffer memory 108. Is possible. When reverse playback progresses and moves to the next GOP, it is necessary to read data for 1 GOP from the disk of the drive 603 again, so that 200 msec is required and playback is interrupted. For example, sufficient buffer memory 608 is provided. By securing a large capacity and pre-reading data for the previous GOP in advance during decoding, the waiting time required for reading can be eliminated. However, since the display of the first frame across the GOP is only possible after decoding a plurality of P pictures, it may take about 100 msec in some cases. In that case, the user feels discontinuity at the boundary of the GOP, but since the time is relatively short, the quality of reproduction is not greatly impaired. In addition, with the evolution of semiconductor processes, the decoding process speed is also dramatically increased, enabling continuous playback without interruption.

  As described above, a P picture as an access point is appropriately provided in the middle of a GOP, attribute information of the P picture in the GOP is provided, and before starting reverse playback, all the P pictures as access points are decoded, By storing the decoded data in the reverse playback frame memory 609 prepared separately from the frame memory of the decoder 607, smooth reverse playback can be performed even in MPEG4-AVC.

6 is a diagram for explaining a configuration of a GOP in MPEG4-AVC according to Embodiment 1. FIG. 3 is a configuration example of a GOP when an access point for a P picture is set in the first embodiment. FIG. 10 is a diagram illustrating a relationship between a video file on an optical disc and management data in Entry_map () according to the first embodiment. 6 is a structure diagram of index information for accessing the GOP head in Embodiment 1. FIG. FIG. 10 is a diagram illustrating a configuration of GOP_Structure () in which attribute information of a P picture in a GOP is described in Embodiment 1. 1 is a diagram illustrating a configuration of a playback device according to Embodiment 1. FIG. 6 is a diagram illustrating a GOP configuration in which P pictures serving as two access points are present in Embodiment 1. FIG.

  1 navigation data, 2 video data, 3 video files, 600 playback device, 601 user interface (I / F) unit, 602 CPU, 603 drive, 604 drive control unit, 605 work memory, 606 system bus, 607 decoder, 608 buffer Memory, 609 reverse playback frame memory, 610 display device.

Claims (2)

The frames after the frame designated as the random access focus do not refer to the frames before the random access point, the random access point is an I picture or a P picture, and the P picture existing in the GOP is the next A video information playback apparatus for playing back a video information recording medium in which attribute information of a P picture indicating whether the access point is a picture necessary for decoding the access point or not is recorded.
A frame memory for holding video information after decoding the frame serving as the access point;
Before starting reverse playback, decode the frame that is the access point included in the GOP,
Holding the decoded video information of the access point in the frame memory;
A video information playback apparatus, wherein video information stored in the frame memory is used as a reference frame for decoding during reverse playback. The frames after the frame designated as the random access focus do not refer to the frames before the random access point, the random access point is an I picture or a P picture, and the P picture existing in the GOP is the next A video information playback method for playing back a video information recording medium in which attribute information of a P picture indicating whether the access point is a picture necessary for decoding the access point or not is recorded.
A frame memory for holding video information after decoding the frame serving as the access point;
Before starting reverse playback, decode the frame that is the access point included in the GOP,
Holding the decoded video information of the access point in the frame memory;
A video information playback method, wherein video information stored in the frame memory is used as a reference frame at the time of decoding during reverse playback.

Images