JP2009296078A - Encoded data reproducing apparatus, encoded data reproducing method, and encoded data reproducing program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To reproduce natural pictures without the feeling of incompatibility in an early stage by generating an appropriate DPB state even when a decoding start point is not an IDR picture when random access is instructed. <P>SOLUTION: The encoded data reproducing apparatus 100 includes: a header information extraction unit 150 for successively extracting header information from the IDR picture positioned immediately before an optional access point; a DPB management unit 160 for deriving a DPB state in the access point through simple DPB control by reference information in the header information from the IDR picture to the access point and executing the DPB control according to the header information with the derived DPB state as a start state after the access point; a picture control unit 166 for generating the decoded data of a picture unit according to the DPB state after the access point; and a picture reproducing unit 140 for reproducing the generated decoded data. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to an encoded data reproducing apparatus, an encoded data reproducing method, and an encoded data reproducing program for reproducing encoded data obtained by encoding a moving image signal in units of pictures through DPB control.

  In H.264 / MPEG-4 AVC (hereinafter referred to as “H.264”), since a reference picture can be arbitrarily selected, it is also possible to refer to a picture preceding an immediately preceding I picture. Therefore, H.I. In H.264, encoded data such as a DPB (Decoded Picture Buffer) state, an FN (Frame Number), and a POC (Picture Order Count) are used so that subsequent pictures can be correctly decoded without using a picture before that picture. An IDR (Instantaneous Decoding Refresh) picture for resetting a state necessary for decoding is prepared. When random access is instructed, the normal IDR picture is reproduced.

  Non-closed GOP in MPEG2 and H.264 In H.264, since it is possible to refer to a picture preceding the previous I picture in the coding order, even if random access using the I picture as an access point is instructed, reference is made to the DPB. There has been a problem that there is no reference picture to be decoded and decoding cannot be started from the I picture.

Therefore, in MPEG2, a technique is disclosed in which a blue background is displayed until the first I picture is decoded and a disordered image is suppressed (for example, Patent Document 1). In addition, when random access is instructed, a technique for replacing the picture in the reference picture list with a decoded picture until the next IDR picture is detected and forcibly starting playback from an arbitrary access point is also disclosed. (For example, Patent Document 2).
JP-A-8-214265 JP 2007-184791 A

  As mentioned above, H.M. In the H.264 encoding method, when an LT (Long Term) picture that can arbitrarily select a reference picture and can be retained as a long-term reference picture is used, decoding is started over a long period until the IDR picture is decoded. A picture before the point may be referred to, and there is a problem that a period in which an image is disturbed is long as compared with MPEG2 encoding.

  Further, when decoding is started from an I picture that is not an IDR picture, there is a discrepancy between the original DPB state and the DPB state when decoding is suddenly started from the I picture. When a memory management control operation (MMCO) command (hereinafter simply referred to as an MMCO command) or a reference picture reordering information for arbitrarily manipulating a picture in a picture is added There was also a problem that the processing could not be performed.

  In view of such a problem, the present invention generates an appropriate DPB state when a random access is instructed even if the decoding start point is not an IDR picture, and reproduces a natural image without a sense of incongruity at an early stage. It is an object of the present invention to provide an encoded data reproduction apparatus, an encoded data reproduction method, and an encoded data reproduction program capable of performing the above.

  In order to solve the above-described problems, a typical configuration of the present invention is an encoded data reproduction apparatus that reproduces encoded data obtained by encoding a moving image signal in units of pictures through DPB control, and provides arbitrary access. A header information extraction unit that sequentially extracts header information from an IDR picture located immediately before the point, and a simple DPB control based on reference information in header information from the IDR picture to the access point, derives a DPB state at the access point, and accesses After the point, a DPB management unit that executes DPB control according to the header information with the derived DPB state as a start state, an image control unit that generates decoded data in units of pictures according to the DPB state after the access point, and the generated decoding And an image playback unit for playing back data.

  Here, the simple DPB control using the reference information (picture type, period identifier, display order, encoding order, reference necessity) in the header information is used for the original DPB state at the access point. Derived DPB control). Therefore, even when an arbitrary I picture other than an IDR picture is designated as an access point, an appropriate DPB state can be generated, a quality degradation of a decoded image can be prevented, and a natural image without a sense of incongruity can be generated. It can be played back. Also, by omitting output control and actual decoding processing (inverse VLC, inverse quantization, inverse DCT, motion compensation, image control) originally executed in DPB control, only simple DPB control based on reference information is performed. The DPB state can be quickly derived, and early image reproduction is possible without delaying the decoding start time.

  When the reference picture included in the DPB state at the access point is referred to, the image control unit may replace the reference picture with an arbitrary picture after the access point for reference. Further, the reference picture may be referred to by replacing it with a predetermined picture.

  In the present invention, since actual decoding is not involved, the arrangement of reference pictures represented in the DPB state is appropriate, but there is no image information obtained by decoding the pictures. Therefore, the image control unit determines whether or not the reference picture included in the DPB state at the access point is referred to, and when referring to it, the decoded data of another adjacent picture or a predetermined picture prepared in advance is used. A predicted image is generated by referring to the image. With such a configuration, it is possible to prevent the decoded image from being disturbed, although a certain degree of image quality degradation occurs.

  Another typical configuration of the present invention is an encoded data reproducing apparatus that reproduces encoded data obtained by encoding a moving image signal in units of pictures through DPB control, and is located immediately before an arbitrary access point. A header information extraction unit that sequentially extracts header information from the IDR picture, a DPB management unit that updates the DPB state through DPB control based on the header information extracted from the IDR picture, and decoding in units of pictures according to the DPB state after the access point An image control unit that generates data and an image reproduction unit that reproduces the generated decoded data are provided.

  Unlike the simple DPB control described above, in the present invention, a complete DPB state including output control can be derived by performing original DPB control from an IDR picture located immediately before an arbitrary access point. Here, since the existing DPB control configuration can be used as it is, an effect can be obtained with certainty without major changes in configuration or program.

  The header information extraction unit may not transmit the header information of the non-reference picture to the DPB management unit between the IDR picture and the access point.

  In the above-described invention, since there is no actual output control of the decoded image up to the access point, it is not necessary to control the non-reference picture. Therefore, the control result can be derived efficiently and quickly by not including the non-reference picture in the simple DPB control or the DPB control.

  Another typical configuration of the present invention is an encoded data reproduction method for reproducing encoded data obtained by encoding a moving image signal in units of pictures through DPB control, and is positioned immediately before an arbitrary access point. The header information from the IDR picture is sequentially extracted, the DPB state at the access point is derived through the simple DPB control based on the reference information in the header information from the IDR picture to the access point, and the derived DPB state is started after the access point As described above, DPB control is executed according to the header information, decoded data in units of pictures is generated according to the DPB state, and the generated decoded data is reproduced.

  Another typical configuration of the present invention is an encoded data reproduction program that causes a computer to function as an encoded data reproduction device that reproduces encoded data obtained by encoding a moving image signal in units of pictures through DPB control. A header information extraction unit that sequentially extracts header information from an IDR picture located immediately before an arbitrary access point, and an access point through simple DPB control using reference information in header information from the IDR picture to the access point. A DPB management unit that derives the DPB state at the access point and executes DPB control according to the header information with the derived DPB state as the start state after the access point, and an image that generates decoded data in units of pictures according to the DPB state after the access point The controller and the generated decryption data Wherein the image reproduction unit for reproducing the data, to the be made to function.

  The components corresponding to the technical idea of the encoded data reproduction device and the description thereof can be applied to the encoded data reproduction method and the encoded data reproduction program.

  As described above, according to the present invention, when a random access such as a track search, a chapter search, or a time code search is instructed, an appropriate DPB state is generated even if the decoding start point is not an IDR picture. It is possible to reproduce a natural image without any sense of incongruity. Therefore, even if the MMCO command and reordering information are added to the encoded data after the access point, the DPB state necessary for the processing is held, so that the processing can be normally performed. It is possible to prevent image distortion due to state mismatch.

  Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. The dimensions, materials, and other specific numerical values shown in the embodiment are merely examples for facilitating understanding of the invention, and do not limit the present invention unless otherwise specified. In the present specification and drawings, elements having substantially the same function and configuration are denoted by the same reference numerals, and redundant description is omitted, and elements not directly related to the present invention are not illustrated. To do.

  In the present embodiment, the problem caused by the DPB state mismatch when the access point is not an IDR picture is solved by forming the original DPB state through simple DPB control from the immediately preceding IDR picture. In this way, a natural image without a sense of incongruity can be reproduced from the access point. Here, the overall configuration of the encoded data reproduction apparatus will be described first.

  FIG. 1 is a block diagram showing the configuration of the encoded data reproduction device 100 according to this embodiment. The encoded data reproduction device 100 includes a disk drive 110, a stream buffer 120, a decoder 130, and an image reproduction unit 140.

  The disk drive 110 has a moving image signal of ITU-T H.264. The encoded data encoded (encoded) in units of pictures using the H.264 encoding method is read from the storage medium 102. Here, a disk-shaped medium is assumed as the storage medium 102, but various media such as a static memory and an HDD can be applied. The stream buffer 120 temporarily holds the encoded data read by the disk drive 110 and outputs the encoded data as needed in accordance with a command from the decoder 130. The decoder 130 decodes (decodes) the encoded data from the stream buffer 120 to generate decoded data. The image reproduction unit 140 reproduces the decoded data generated by the decoder 130 and transmits the reproduction signal to the monitor 104.

  The decoder 130 includes a header information extraction unit 150, an inverse VLC unit 152, an inverse quantization unit 154, an inverse DCT unit 156, an adder 158, a DPB management unit 160, and a DPB state holding unit 162. A picture list creation unit 164, an image control unit 166, a frame memory 168, and a motion compensation unit 170 are configured.

  The header information extraction unit 150 sequentially extracts header information of each picture in the encoded data from the stream buffer 120, and supplies the header information to the inverse VLC unit 152, the DPB management unit 160, and the reference picture list creation unit 164. In this embodiment, the position where header information is extracted is an IDR picture located immediately before an arbitrary access point. However, when the access point is an IDR picture, the original DPB control may be executed from the IDR picture without performing the simple DPB control described later.

  Further, the header information extraction unit 150 decodes from the previous IDR picture to the access point into an inverse VLC unit 152, an inverse quantization unit 154, an inverse DCT unit 156, an adder 158, an image control unit 166, and a frame memory 168, which will be described later. Outputs flag data to indicate that no decoding process is required. Then, the access point outputs decoding flag data indicating that decoding processing is necessary, starts decoding processing, and outputs a stream to the inverse VLC unit 152.

  Furthermore, if the picture is a non-reference picture from the IDR picture to the access point, the header information extraction unit 150 does not transmit the header information to the DPB management unit 160. The simple DPB control of this embodiment does not involve the actual output control of the decoded image up to the access point, so there is no need to control the non-reference picture. Therefore, the control result can be derived efficiently and quickly by not including the non-reference picture in the simple DPB control or the DPB control.

  The inverse VLC unit 152 performs inverse VLC processing on the encoded data following the header information using the header information supplied from the header information extraction unit 150, and inversely quantizes the inverse VLC processed data and the quantization step size. The motion vector information is output to the motion compensation unit 170 described later. The inverse quantization unit 154 inversely quantizes the input data according to the quantization step size designated by the inverse VLC unit 152 and outputs the data to the inverse DCT unit 156. The inverse DCT unit 156 performs inverse DCT processing on the input data and supplies it to the adder 158. The adder 158 adds the output of the inverse DCT unit 156 and the output of the motion compensation unit 170 and outputs the result to the image control unit 166.

  On the other hand, the DPB management unit 160 performs the DPB control through the DPB state held in the DPB state holding unit 162, for example, the arrangement of reference pictures and the parameters associated therewith (picture type, period identifier, display order, encoding order, Reference necessity / output state) is updated, and the DPB state is supplied to the image control unit 166 and the reference picture list creation unit 164. Here, the DPB state holding unit 162 that holds the DPB state and the DPB of the frame memory 168 that stores the actual decoded data according to the DPB state are shown as separate configurations, but they can also be controlled collectively.

  In this embodiment, the DPB management unit 160 particularly references information (picture type, period identifier, display order, encoding order, reference necessity) of header information from the IDR picture to the access point extracted by the header information extraction unit 150. The DPB state at the access point is derived through the simple DPB control according to), and the original DPB control is executed after the access point with the derived DPB state as the start state. Here, the simplified DPB control refers to DPB control specialized only for the arrangement of reference pictures.

  Therefore, even when an arbitrary I picture other than an IDR picture is designated as an access point, an appropriate DPB state can be generated, a quality degradation of a decoded image can be prevented, and a natural image without a sense of incongruity can be generated. It can be played back. Also, by omitting output control and actual decoding processing (inverse VLC, inverse quantization, inverse DCT, motion compensation, image control) originally executed in DPB control, only simple DPB control based on reference information is performed. The DPB state can be quickly derived, and early image reproduction is possible without delaying the decoding start time.

  The reference picture list creation unit 164 creates a reference picture list using the header information extracted by the header information extraction unit 150 and a command from the DPB management unit 160 and supplies the reference picture list to the image control unit 166. The reference picture list includes a description that allows the motion compensation unit 170 to specify the reference picture at which position of the DPB, for example, the DPB arrangement number, the address of the frame memory 168, and FN (LTPN (LT Picture Number)). Etc. are described.

  The image control unit 166 instructs the output of the decoded image and the storage to the frame memory 168 according to the DPB state held in the DPB state holding unit 162, and the reference picture supplied from the reference picture list creation unit 164 According to the list, the frame memory 168 is instructed to output the reference picture to the motion compensation unit. The frame memory 168 holds the decoded image (decoded data), and outputs or deletes the held image in accordance with an instruction from the image control unit 166. The motion compensation unit 170 acquires the reference picture supplied from the frame memory 168 according to the reference picture list created by the reference picture list creation unit 164, performs motion compensation processing, and outputs the reference picture to the adder 158.

  Hereinafter, in order to facilitate understanding of the present embodiment, first, normal DPB control after the access point will be described, and then, a characteristic DPB state forming operation in the present embodiment will be described.

  As described above, H.P. In H.264, it is possible to refer to arbitrarily distant pictures. Therefore, the decoded image is stored in the frame memory 168 as a DPB for management. At this time, the DPB management unit 160 causes the DPB state holding unit 162 to hold the arrangement of the reference picture to be held and the parameters thereof as the DPB state, and update it according to the header information.

  FIG. 2 is an explanatory diagram for describing DPB control in H.264. FIG. 2A shows the display order of each picture, and FIG. 2B shows the decoding order. In FIG. 2 and subsequent figures, “IDR” indicates an IDR picture, “I” indicates an I picture, “P” indicates a P picture, “B” indicates a B picture, and the accompanying numerical values are mainly displayed in display order (POC). ), And arrows indicate reference relationships.

  Assuming a data stream to be reproduced from an IDR picture as shown in FIG. 2A, the DPB state is reset at the start of decoding. H. Under H.264, for example, the P picture P6 can refer to the IDR picture IDR0 further apart in time, not the immediately preceding P picture P3. However, the previous picture cannot be referenced across the IDR picture IDR0. As for the B picture, a P picture P6 that is separated in time, such as a B picture B1, is referred to, or P pictures P3 and P6 that are located in a time position, such as a B picture B2, are referred to. It is possible to refer to the B picture B4 as in B5.

  When decoding the data stream shown in FIG. 2 (a), pictures that each picture refers to when decoding must first be decoded. Therefore, as shown in FIG. 2B, P pictures P3 and P6 are decoded using IDR picture IDR0 having no reference picture, and B pictures B1, B2, B4, B5 is decoded. Decoding is performed in the same procedure from P picture P9.

  FIG. 3 is a conceptual diagram illustrating the DPB management procedure of the DPB management unit 160 for the data stream of FIG. Here, the DPB management unit 160 manages the DPB state, that is, the arrangement of reference pictures to be included in the DPB and the parameters associated therewith (picture type, period identifier, display order, encoding order, reference necessity, output state). . Decoded data obtained by actually decoding each picture is held in the DPB in the frame memory 168.

  As described above, parameters (picture type, period identifier, display order, encoding order, reference necessity, output state) are attached to each picture. Here, the picture type is identification of an I picture, P picture, and B picture, the period identifier is an identifier of a short-term reference picture (ST picture) and a long-term reference picture (LT picture) described later, the display order is POC, and the coding order Indicates whether FN is used as a reference picture by used (used for reference) or unused (unused for reference), and the output state is need (needed for output) or not needed (not needed to output) Indicates whether output is waiting or output. Here, a case where the period identifier is ST will be described.

  In FIG. 3A, the DPB state is reset by the IDR picture IDR0, and the leading IDR picture IDR0 is set as the current picture. The reference picture forms a DPB state at any time. When the current picture to be decoded is P picture P6, the header information of IDR picture IDR0 and P picture P3 is held in the first and second areas in the DPB state as shown in FIG. Nothing is held in the second area. Here, the header information of the reference picture is managed through each holding area in such a DPB state.

  FIG. 4 is a diagram for explaining POC and FN assigned to the picture of FIG. If the previous picture is a reference picture, the value of FN is incremented. For example, if the sequence is P picture, B picture, B picture, P picture, FN becomes N (integer), N + 1, N + 1, N + 1. Therefore, the FN of the data stream as shown in FIG. 4 is 0, 1, 2, 3, 3, 3, 4, 4, 5, 5.

  FIG. 5 is a conceptual diagram for explaining the DPB management procedure of the DPB management unit 160 for the data streams after FIG. The marking process of a picture included in the DPB state includes a sliding window method in which information is not particularly added and an MMCO method in which information called an MMCO command is added. In particular, when a command is not added, a sliding window is used. A method is used. Here, the sliding window type marking process will be described.

  In FIG. 5, it is assumed that the upper limit (MaxDPB) of the number of frames in the DPB state is 3, and the maximum number of stored reference pictures (no_of_ref) is 3. FIG. 5A shows the state of the DPB after decoding P6 in FIG. Since pictures IDR0, P3, and P6 in the DPB state are all reference pictures, the necessity of reference is marked “used”. Since the IDR picture IDR0 has already been output, the output state is marked “not needed”, and the P pictures P3 and P6 are marked “needed” because they are waiting to be output. When the B picture B1 of FN3 is subsequently decoded, the “unused” is marked because the B picture B1 is not a reference picture.

  Here, comparing the POC of the picture in the DPB state with the POC of the B picture B1 in the current picture, the POC of the B picture B1 is the smallest among the “needed” marks, so the decoded data of the picture B1 is output. And transferred to the output buffer. In this case, the B picture B1 becomes “unused” and “not needed” and does not need to be held in the DPB state. Therefore, the actual DPB does not hold the decoded data. Since the B picture B2 is also output in the same manner, the DPB state remains as shown in FIG.

  Subsequently, as shown in FIG. 5B, when the B picture B4 is decoded, the B picture B4 is marked as “used” because it is a reference picture. Accordingly, the number of reference pictures marked “used” is 4, which exceeds the maximum number of stored pictures, so the picture with the smallest FN, that is, the IDR picture IDR0 is marked “unused”. Then, the IDR picture IDR0 becomes “unused” and “not needed”, and is deleted from the DPB state because it becomes unnecessary in the subsequent decoding process. In the DPB state, pictures P3, P6, and B4 are left.

  Following the B picture B4, the B picture B5 having the same FN is decoded. Since the B picture B5 is not a reference picture, “unused” is marked as shown in FIG. Next, the POC of the picture held in the DPB state is compared with the POC of the B picture B5. Since the POC of B5 is not the minimum, “needed” is marked at this point. However, in the DPB state, since the number of reference pictures has reached the maximum storage number, the picture with the smallest POC, that is, the P picture P3 is output and marked as “not needed”. However, since the P picture P3 is marked “used”, it cannot be deleted from the DPB state.

  Therefore, next, the picture with the smallest POC, that is, the B picture B4 is output and marked as “not needed”. However, since the B picture B4 is also marked “used”, it cannot be deleted from the DPB state. Since the picture with the next smallest POC is the B picture B5, the B picture B5 output from the B picture B5 and marked “not needed” becomes “unused” and “not needed” and is deleted without being held in the DPB state. Is done.

  In this way, the DPB state is sequentially updated, and the arrangement of the reference picture is retained. When the next IDR picture is decoded, all the pictures in the DPB state are marked as “unused”, all the pictures are output in the POC order, deleted from the DPB state holding unit 162, and the DPB state is empty. Become. Accordingly, no decoded data remains in the DPB itself. At this time, it is possible not to output the picture according to the information in the encoded data. In this way, H.C. In H.264, it is possible to refer to a picture at a distant position, but it is not possible to refer to a previous picture across IDR pictures. POC and FN are reset to 0 when an IDR picture is detected. From this background, H.C. In H.264, decoding has been started from the immediately following IDR picture regardless of the position of the access point.

  Next, LT picture and MMCO marking will be described. As described above, in the marking process of the sliding window method, the FNs of the pictures stored in the DPB state are compared, and “unused” is marked in order from the pictures with the smallest FN. In contrast, in the MMCO marking method to which the MMCO command is added, an arbitrary picture can be set to “unused” regardless of the size of the FN.

  Also, by using the MMCO command, the ST picture in the DPB can be changed to the LT picture. Here, the LT picture is a picture that stays in the DPB as a reference picture for a long time. Unlike the ST picture, the LT picture is not changed to “unused” in the sliding window method. An LT picture is marked as “unused” when an MMCO command for marking the LT picture as “unused” is added, an IDR picture is decoded, or an MMCO command that performs the same operation as IDR Is added.

  When LT pictures are used in this way, pictures can be stored in the DPB for a long period of time regardless of the encoding order, and can be used as a reference picture for a long period of time. The sliding window method and the MMCO method can be switched for each coding unit including a plurality of macroblocks called slices.

  FIG. 6 is an explanatory diagram for explaining DPB processing when MMCO marking and an LT picture are used. For example, the MMCO command “allocate LTPN0 to the picture of PN (Picture Number) 1” and the MMCO command “assign LTPN1 to the current picture” are added to the encoded data of the P picture P6 in FIG. To do.

  Here, PN takes into account the field order based on FN and can be handled in the same way as FN. LTPN is a number for identifying an LT picture. In this case, the state of the DPB after decoding the P picture P6 is as shown in FIG. Of the pictures stored in the DPB, the P pictures P3 and P6 are LT pictures and are assigned LTPN. Subsequently, the B pictures B1, B2, B4, and B5 are decoded. The DPB processing at this time is performed in the same manner as in the description of the DPB processing with reference to FIGS. FIG. 6B shows the state of the DPB after decoding B5.

  When the decoding of the B picture B5 is completed, the P picture P9 is then decoded. Since the P picture P9 is a reference picture, it is marked “used”. Here, the number of reference pictures marked “used” is 4, which exceeds the maximum number of stored pictures. Therefore, the picture with the smallest FN must be “unused”, but the P picture P3 with the smallest FN is an LT picture and is not marked “unused” in the sliding window type marking. Similarly, since the P picture P6 is also an LT picture, it is not marked “unused”, but instead, the B picture B4 is marked “unused”.

  Since the B picture B4 becomes “unused” and “not needed”, the B picture B4 is deleted from the DPB state, and the P picture P9 is stored in the DPB state holding unit 162 so as to be added to the DPB state. Thus, the DPB state of the DPB state holding unit 162 becomes the state shown in FIG. In this way, by using the LT picture, the picture that is marked “unused” in the sliding window method can be maintained in the “used” state, and a later picture, for example, B in FIG. In decoding of the picture B7, the LT picture can be used as a reference picture.

  Next, the reference picture list will be described. The reference picture list is a list in which the pictures marked “used” among the header information of the reference pictures held in the DPB state holding unit 162 are arranged in order. A reference picture is designated by adding information indicating what number of picture in the list to each macro block (MB) or sub MB.

  In a P picture, first, the LT pictures of DPB are arranged in ascending order of LTPN. In the B picture, in the first reference picture list, ST pictures having a POC smaller than that of the picture to be decoded are arranged in order of increasing POC, and thereafter ST pictures having a POC larger than that of the picture to be decoded are arranged in order of increasing POC. Thereafter, the LT pictures are arranged in ascending order of LTPN. In another reference picture list, ST pictures having a POC larger than that of the picture to be decoded are arranged in the order of increasing POC, ST pictures having a POC smaller than that of the picture to be decoded are arranged in order of increasing POC, and then the LT picture Are arranged in ascending order of LTPN. For example, in the example of FIG. 6, since the DPB state when decoding the P picture P9 is the state of FIG. 6B, the reference picture list is “B4, P3, P6”.

  The reference picture list can be rearranged by sending reordering information in a slice header added for each slice. The reordering information includes the position in the reference picture list and the FN of the reference picture to be inserted at the position. By using this, an arbitrary picture can be placed at an arbitrary position in the reference picture list, or a plurality of the same pictures can be stored. It becomes possible to put.

  The DPB control described above assumes that decoding is started from the IDR picture immediately after the access point. In this embodiment, as shown below, when random access is instructed, simple DPB control with a light load is performed from the IDR picture immediately before the access point, and the DPB state that should be originally generated is generated to reproduce from the access point. Is possible.

  FIG. 7 is an explanatory diagram for explaining the simple DPB control before the access point. The data stream shown in FIG. 7 is an excerpt of part of the encoded data, and it is assumed that the encoded data exists before and after this. The POC and FN of the encoded data are as shown in FIG. 7, and the access point is an I picture I6 between two IDR0 pictures.

  The header information extraction unit 150 detects the header of the IDR picture located immediately before the access point according to the access point designated from the outside from the encoded data from the stream buffer 120. Here, it is assumed that the access point is designated from the outside. However, the present invention is not limited to this, and means for designating the access point may be provided in the encoded data reproduction device 100 itself. In addition, the header information extraction unit 150 outputs the decryption flag data to indicate that the decryption process is unnecessary.

  Furthermore, the header information extraction unit 150 extracts header information from the encoded data supplied from the stream buffer 120. From the reference necessity information included in the extracted header information, it is determined whether or not the picture is a reference picture. If the picture is not a reference picture, the picture is ignored and the processing proceeds to the next picture. . In addition, the header information extraction unit 150 does not supply the encoded stream to the inverse VLC unit 152 from the previous IDR picture to the access point, and does not execute decoding processing such as inverse VLC, inverse quantization, and inverse DCT. Further, the image control unit 166 does not perform processing such as image output and storage.

  When the target picture is a reference picture, the header information extraction unit 150 supplies the extracted header information together with the decoding flag data to the DPB management unit 160 and the decoding flag data to the image control unit 166. If the supplied decryption flag data indicates that the decryption process is not necessary, the DPB management unit 160 performs simple DPB control instead of normal DPB control. In the simple DPB control, among the conventional DPB controls, control related to image output, that is, management of “needed” or “not needed” is not always performed, and “not needed” is always marked, and control related to a reference picture, ie, “ Only “used” or “unused” is managed. Therefore, the DPB control after the access point described above is followed except that only the DPB state is generated and control regarding image output is not performed.

  Then, after the access point, the DPB management unit 160 executes the original DPB control according to the header information with the DPB state derived in this way as a start state.

  FIG. 8 is an explanatory diagram for explaining the DPB state at the access point when the simple DPB control is executed. In particular, FIG. 8A shows a DPB state, and FIG. 8B shows encoded data from an access point. Here, it is assumed that the access point is the I picture I6, the upper limit of the number of DPB pictures (frames) is, for example, 2 pictures, and the maximum number of reference pictures stored is 2 pictures. Note that the encoded data shown in FIG. 8 shows only the first part after the detected header, and does not describe the sequence parameter set or the like.

  As described above, before starting the decoding process of the I picture I6 serving as the access point, header information from the IDR picture located immediately before the access point to the access point is extracted, and when the picture is a reference picture, DPB is extracted. Since the simple control is executed, the marking regarding the reference of the picture in the DPB and the information of the FN and POC are in the same state as when the conventional decoding process is performed from the IDR picture located immediately before the access point. Yes. However, since “not needed” is forcibly written in the output state, there is a possibility that the output state is different from the original DPB state.

  When the header information extraction unit 150 detects the header position corresponding to the access point, the header information extraction unit 150 transmits decoding flag data indicating that decoding processing is necessary to each component, and transmits the subsequent encoded stream to the inverse VLC unit 152. Then, decoding processing such as inverse VLC, inverse quantization, and inverse DCT is performed by a normal method. The DPB management unit 160 also performs DPB control by a normal method according to the supplied header information.

  The image control unit 166 operates the decoded data in the frame memory 168 according to the DPB state input from the DPB management unit 160, and sequentially generates decoded data.

  At this time, the reference picture list creation unit 164 creates a reference picture list according to the DPB state input from the DPB management unit 160.

  FIG. 9 is an explanatory diagram for explaining DPB control immediately after the access point. For example, as shown in FIG. 9A, a reference picture list used when decoding an encoded B picture B5 subsequent to an I picture I6 includes a P picture P4 in idx0 and an I picture I6 in idx1. The control unit 166 instructs the frame memory 168 to supply the reference picture specified in FIG. 9A to the motion compensation unit 170 according to the reference picture list. Here, the image control unit 166 refers to the DPB state holding unit 162 and determines whether the reference picture referred to by the motion compensation unit 170 is included in the DPB state at the access point. Is replaced with an arbitrary picture after the access point. Further, the reference picture may be referred to by replacing it with a predetermined picture.

  For example, when the reference picture list as shown in FIG. 9A is supplied, the I picture I6 exists in the DPB state of the DPB state holding unit 162, and the decoded data is also stored in the frame memory 168. However, although the P picture P4 exists in the DPB state of the DPB state holding unit 162, the decoded data is not stored in the frame memory 168. Here, for example, the decoded I picture I6 is replaced at the position P4 in the frame memory 168, and the decoding process of the B picture B5 is performed.

  In the present embodiment, since actual decoding is not involved, the arrangement of the reference picture represented in the DPB state is appropriate, but there is no image information obtained by decoding the picture. At this time, some decoded picture is used instead of the original image information. Therefore, the image control unit 166 determines whether or not the reference picture included in the DPB state at the access point is referred to, and in the case of referring, other adjacent pictures, for example, pictures available at that time Among them, a picture that is similar to the original picture and that is closest in time or simply located in the reference list, or a predetermined picture that is prepared in advance, such as “pure white”, “pure black”, or “gray” A predicted image is generated by adaptively referring to the decoded data. With such a configuration, it is possible to prevent the decoded image from being disturbed, although a certain degree of image quality degradation occurs.

  FIG. 9B shows, as a comparative example, a reference picture list and a DPB state when decoding is started from an I picture I6 without applying this embodiment. If the B picture B5 is to be decoded under such circumstances, only the I picture I6 is held as the DPB state, and a reference picture list as shown in FIG. 9B is generated. In this case, although the picture to be decoded is a B picture that requires two images, there is no picture corresponding to idx1 in the figure, which may cause a problem in the decoding process.

  In this embodiment, since simple DPB control is performed, even when an I picture other than IDR is designated as an access point in this way, in the decoding of encoded data after the access point, inconsistency with the DPB state The decoding process can be performed without the occurrence of. In addition, since the marking state related to the DPB reference at the access point is maintained in the same manner as when decoding processing is performed from the previous IDR picture, reordering information and MMCO information are used in the encoded data after the access point. Can be processed without inconsistency.

  Further, since the decoding process is not performed on the encoded data from the immediately preceding IDR picture to the access point, the DPB control process is performed by a simpler method than the normal DPB control process only for the reference picture. The decoding of the encoded data after the access point can be started promptly.

(Encoded data reproduction program)
FIG. 10 is a functional block diagram showing a typical example of a computer (information processing apparatus) 200 capable of reproducing encoded data by the encoded data reproducing apparatus 100 according to this embodiment. The computer 200 includes a central processing unit 210, a temporary storage device 212, an external storage device 214, a communication unit 216, an input unit 218, and an output unit 220.

  A central processing unit (CPU) 210 manages and controls the entire computer 200 using programs and applications in the temporary storage device 212 and the external storage device 214. The temporary storage device 212 includes a ROM, a RAM, an EEPROM, a nonvolatile RAM, and the like, and temporarily stores a program processed by the central processing unit 210, encoded data, and the like. The external storage device 214 includes a flash memory, an HDD, and the like, and stores programs processed by the central processing unit 210, encoded data, and the like. The communication unit 216 is connected to various electronic devices and servers via the communication line 230, and can transmit / receive video data and encoded data to / from them. The input unit 218 can input encoded data from the storage medium 102. The output unit 220 can output video output to the monitor 104 or the like.

  The reproduction of the encoded data described above is performed by the central processing unit 210 executing a program. Therefore, at the same time as the encoded data reproduction apparatus 100 is provided, an encoded data reproduction program that causes the computer 200 to function as the encoded data reproduction apparatus 100 is also provided. The encoded data reproduction program may be read from a recording medium and loaded into a computer, or may be transmitted via a communication line 230 and loaded into a computer.

(Encoded data reproduction method)
Also provided is an encoded data reproduction method for reproducing encoded data obtained by encoding a moving image signal in units of pictures through DPB control.

  FIG. 11 is a flowchart showing specific processing of the encoded data reproduction method. When an arbitrary access point by random access is specified, first, the header information extraction unit 150 detects the position of the header of the IDR picture located immediately before the arbitrary access point in the encoded data from the stream buffer 120, and Header information from the IDR picture is sequentially extracted (S300).

  Subsequently, it is determined whether or not the picture for which header information is to be extracted has reached the access point (S302). If the picture has not arrived ("NO" in S302), the DPB management unit 160 uses only the reference information in the header information. The simple DPB control is performed to derive the DPB state at the access point (S304). Such simple DPB control is repeated until the access point is reached.

  When the access point is reached (“YES” in S302), that is, after the access point, the DPB management unit 160 extracts the DPB state derived by the simplified DPB control as the start state, and the header information extraction unit 150 extracts it. The DPB state is derived according to all the header information (S306), and the image control unit 166 generates decoded data for each picture according to the DPB state (S308). Then, the image reproduction unit 140 reproduces the decoded data generated in this way (S310). Finally, it is determined whether or not a reproduction end command has been issued (S312), and the composite data reproduction by the original DPB control is repeated until the reproduction end command is issued ("YES" in S312).

  Even in this encoded data reproduction method, when random access is instructed, it is possible to generate an appropriate DPB state and reproduce a natural image without a sense of incongruity at an early stage even if the decoding start point is not an IDR picture. Become. Therefore, even if the MMCO command and reordering information are added to the encoded data after the access point, the DPB state necessary for the processing is held, so that the processing can be normally performed. It is possible to prevent image distortion due to state mismatch.

(Other embodiments)
In the above-described embodiment, the simple DPB control is performed to form the DPB state at the access point, but the original DPB control (full DPB control) is performed from the IDR picture located immediately before the arbitrary access point. Thus, a complete DPB state including output control can be derived.

  Here, unlike the embodiment described above, the DPB management unit 160 updates the DPB state through the original DPB control based on the header information extracted from the IDR picture, and the image control unit 166 also performs the DPB after the access point. Decoded data in units of pictures is generated according to the state. In the present embodiment, since the existing DPB control configuration can be used as it is, an effect can be obtained with certainty without a major program change.

  As mentioned above, although preferred embodiment of this invention was described referring an accompanying drawing, it cannot be overemphasized that this invention is not limited to this embodiment. It will be apparent to those skilled in the art that various changes and modifications can be made within the scope of the claims, and these are naturally within the technical scope of the present invention. Understood.

  In the above embodiment, ITU-T H.264 is used. The encoded data encoded by the H.264 encoding method has been described as an object to be reproduced. However, the present embodiment is not limited to this, and the encoded data encoded by another encoding method may also be an object. Of course.

  Each step in the encoded data reproduction method of the present specification does not necessarily have to be processed in time series in the order described as the flowchart, and may include processing in parallel or by a subroutine.

  INDUSTRIAL APPLICABILITY The present invention can be used for an encoded data reproduction device, an encoded data reproduction method, and an encoded data reproduction program for reproducing encoded data obtained by encoding a moving image signal in units of pictures through DPB control.

It is a block diagram which shows the structure of an encoding data reproduction apparatus. H. 2 is an explanatory diagram for describing DPB control in H.264. FIG. It is a conceptual diagram explaining the DPB management procedure of the DPB management part with respect to the data stream of FIG.2 (b). It is a figure explaining POC and FN given to the picture of Drawing 2 (b). It is a conceptual diagram explaining the DPB management procedure of the DPB management part with respect to the data stream after FIG.3 (b). It is explanatory drawing for demonstrating the process of DPB at the time of using the marking of a MMCO system, and a LT picture. It is explanatory drawing for demonstrating the simple DPB control before an access point. It is explanatory drawing for demonstrating the DPB state in the access point at the time of performing simple DPB control. It is explanatory drawing for demonstrating DPB control immediately after an access point. It is the functional block diagram which showed the typical example of the computer (information processing apparatus) which can reproduce | regenerate encoded data with the encoded data reproduction | regeneration apparatus in this embodiment. It is the flowchart which showed the specific process of the encoding data reproduction method.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 ... Encoded data reproduction apparatus 140 ... Image reproduction part 150 ... Header information extraction part 160 ... DPB management part 166 ... Image control part

Claims (7)

An encoded data reproduction device for reproducing encoded data obtained by encoding a moving image signal in units of pictures through DPB control,
A header information extraction unit that sequentially extracts header information from an IDR picture located immediately before an arbitrary access point;
The DPB state at the access point is derived through simple DPB control based on the reference information in the header information from the IDR picture to the access point, and after this access point, DPB control is performed according to the header information using the derived DPB state as a start state. A DPB management unit for executing
After the access point, according to the DPB state, an image control unit that generates decoded data in units of pictures;
An image reproducing unit for reproducing the generated decoded data;
An encoded data reproducing apparatus comprising:   2. The image control unit according to claim 1, wherein when a reference picture included in a DPB state at the access point is referred to, the reference picture is referred to by replacing the reference picture with an arbitrary picture after the access point. The encoded data reproduction device described in 1.   2. The encoded data according to claim 1, wherein, when a reference picture included in a DPB state at the access point is referred to, the image control unit replaces the reference picture with a predetermined picture for reference. Playback device. An encoded data reproduction device for reproducing encoded data obtained by encoding a moving image signal in units of pictures through DPB control,
A header information extraction unit that sequentially extracts header information from an IDR picture located immediately before an arbitrary access point;
A DPB management unit that updates a DPB state through DPB control using the header information extracted from the IDR picture;
After the access point, according to the DPB state, an image control unit that generates decoded data in units of pictures;
An image reproducing unit for reproducing the generated decoded data;
An encoded data reproducing apparatus comprising:   The code according to any one of claims 1 to 4, wherein the header information extraction unit does not transmit header information of a non-reference picture to the DPB management unit between the IDR picture and an access point. Data reproduction device. An encoded data reproduction method for reproducing encoded data obtained by encoding a moving image signal in units of pictures through DPB control,
Sequentially extract header information from an IDR picture located immediately before an arbitrary access point;
Deriving the DPB state at the access point through simple DPB control with reference information in the header information from the IDR picture to the access point;
After the access point, execute the DPB control according to the header information with the derived DPB state as a start state, generate decoded data in units of pictures according to the DPB state,
An encoded data reproduction method, wherein the generated decoded data is reproduced. Computer
An encoded data reproduction program that functions as an encoded data reproduction device that reproduces encoded data obtained by encoding a moving image signal in units of pictures through DPB control,
The computer,
A header information extraction unit that sequentially extracts header information from an IDR picture located immediately before an arbitrary access point;
The DPB state at the access point is derived through simple DPB control based on the reference information in the header information from the IDR picture to the access point, and after this access point, DPB control is performed according to the header information using the derived DPB state as a start state. A DPB management unit for executing
After the access point, according to the DPB state, an image control unit that generates decoded data in units of pictures;
An image reproducing unit for reproducing the generated decoded data;
An encoded data reproduction program characterized by being made to function.

Images