JP2002374537A - Decoder and its method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To immediately start decoding of a MPEG 2 video bit stream and to enhance the performance for error restoration and a trick play. SOLUTION: A buffer stores an input bit stream and also stores header information and slice information of each layer being a processing layer or more. The slice information includes a start position of each slice on a screen and an address of the data in the buffer. Picture information in the header information includes link information to the slice information of the slices configuring the picture. Prior to decoding the MPEG 2 video bit stream, the header information is analyzed. On the occurrence of an error at variable length demodulation, the range of interpolation is obtained on the basis of the slice information to conduct interpolation.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention
2 (Moving Picture Coding Experts Group) The present invention relates to a decoding apparatus and method suitable for decoding a bit stream of a video.

[0002]

2. Description of the Related Art Digital BS (Broadcast Satellite)
) Broadcasting and digital CS (Communication Satellite)
In the broadcast, video data is compressed by the MPEG2 system, packetized, and stored in an MPEG2-TS (Transport Std.).
ream), and is broadcast on a predetermined carrier.

In MPEG2, motion-compensated predictive coding and D
Video data is compression-encoded by CT (Discrete Cosine Transform).

That is, when image data is converted into frequency domain data by, for example, DCT, the data is biased toward a lower frequency side. Then, when quantizing the DCT-transformed data, the data amount can be reduced by allocating a smaller number of bits to the data on the high frequency side.

[0005] A moving image is composed of a plurality of frames or fields. There is correlation between adjacent frames and fields. For this reason, when the correlation between adjacent frames or fields is strong, the data amount can be reduced by encoding the difference between the image of the adjacent frame or field and the image of the current frame or field. If a motion vector is detected in the screen and motion is corrected using the motion vector, the accuracy of prediction is improved and the data amount can be further reduced.

[0006] In MPEG2, image data is subjected to DCT conversion and converted to frequency domain data, so that the data is compression-coded and the difference between the motion-compensated adjacent frame or field is encoded. Thus, the data is compression-encoded.

In MPEG2, a coding unit consisting of a frame or a field is called a picture, and there are three types of pictures: an I picture, a P picture, and a B picture. I-pictures are coded in the same frame or field. The P picture is obtained by performing one-way inter-frame prediction from a past frame or field and encoding the difference. The B picture is obtained by performing bidirectional inter-frame prediction from two past and future frames and encoding the difference. And
MPEG2 is premised on being applied to a wide range of applications in areas such as broadcasting, communication, and storage media. Therefore, MPEG2 can perform field-unit encoding that is advantageous for interlaced images.

As shown in FIGS. 8 and 9, an MPEG2 video stream has a sequence layer, a GOP (Group Of
Picture) layer, picture layer, slice layer, macroblock layer, and block layer.

[0009] The sequence layer is a hierarchy located at the highest level.
Equivalent to the entire video source. The sequence layer starts with a sequence header and ends with a sequence end.
The sequence header includes information such as the start code of the sequence layer, the number of pixels in the horizontal and vertical directions of the image, and the aspect ratio.

[0010] The GOP layer is a group of pictures including an I picture at the head. This GOP is used at the time of variable speed reproduction or random access. The GOP layer starts with a GOP header and usually consists of 10 to 15 pictures. The GOP header includes information such as a GOP start code and time from the beginning of the sequence.

The picture layer corresponds to one screen, and
It consists of a luminance signal and a chrominance signal of the 2: 0 coding method or the 4: 2: 2 coding method. As described above, pictures include an I picture, a P picture, and a B picture. The picture layer starts with a picture header. The picture header includes information such as a picture layer start code, a picture type, and a quantization matrix.

The slice layer indicates a region divided into a band on the screen. The slice layer includes one or more macroblocks, and is arranged in a raster scanning order from left to right and from top to bottom. The slice layer starts with a slice header. The slice header includes a slice layer start code,
Information indicating the position of the slice is included.

The macroblock layer is a block used as a unit for motion compensation. Macroblocks differ between the 4: 2: 0 coding scheme and the 4: 2: 2 coding scheme. In the case of the 4: 2: 0 coding method, the block includes four blocks of the luminance signal Y and two blocks of the color difference signals Cb and Cr at the corresponding positions. In the case of the 4: 2: 2 coding method, four blocks of the luminance signal Y, two blocks of the color difference signal Cb at the corresponding positions, and Cr
Consists of two blocks. The macroblock layer starts with a macroblock header. The macro block header includes an escape code of the macro block address, the value of the motion vector of the horizontal and vertical components, and the like.
The values of the motion vectors of the horizontal and vertical components are variable-length coded.

The block layer is composed of (8 pixels × 8 lines) data of a luminance signal or a color difference signal. This block layer is a coding unit of DCT. The data of the block layer is composed of each DCT coefficient, and the data of the block layer is variable-length coded in run length.

As described above, the MPEG2 video bit stream has a hierarchical structure, and a header is provided at the head of each layer. When decoding the bit stream of the MPEG2 video, the information of the header of each of these layers is required.

For this reason, conventionally, when decoding an MPEG2 video stream, it waits for the arrival of these headers, and when the information of the header comes, the decoder is set based on the information of the header. Decodes video data.

[0017]

As described above, conventionally, when decoding a bit stream of an MPEG2 video, it is necessary to wait for the header of the bit stream to be decoded before decoding starts immediately. There is a problem that can not be.

Further, since the bit stream cannot be decoded immediately, there is a problem that the performance is reduced during error recovery processing and trick play. Furthermore, since block data and motion vectors are variable-length coded, if an error is detected,
There is a danger that the beginning of the code will be unknown and a runaway will occur.

Therefore, an object of the present invention is to provide an MPEG
An object of the present invention is to provide a decoding apparatus and a decoding method capable of immediately starting decoding of a two-video bit stream.

It is another object of the present invention to provide a decoding apparatus and method capable of improving performance during error recovery processing and trick play.

[0021]

SUMMARY OF THE INVENTION The present invention provides a means for storing an input bit stream, and extracting information contained in a header of a layer higher than a picture layer from the input bit stream to extract header information. Means for generating and accumulating the generated header information, extracting information included in the header of the slice layer from the input bit stream, and determining the position where the bit stream at which the slice starts is stored; Means for generating slice information including a position on the screen to start, and accumulating slice information;
Further, a means for adding link information indicating the position of slice information constituting a picture to the picture information of the header information, a means for reading out the stored header information before starting decoding and analyzing the header information in advance, Means for reading the stored bit stream and decoding the read bit stream using pre-analyzed header information.

The present invention accumulates an input bit stream, extracts information contained in a header of a layer higher than a picture layer from the input bit stream, generates header information, and generates the header information. The information is accumulated, and the information included in the header of the slice layer is extracted from the input bit stream. The position where the bit stream at which the slice starts is stored and the position on the screen at which the slice starts are determined. Generate slice information including the slice information, accumulate the slice information, add link information indicating the position of the slice information constituting the picture to the picture information of the header information, and start decoding the accumulated header information. Before reading, the header information is analyzed in advance, the stored bit stream is read, and the read bit stream is read. A decoding method so as to decode with the previously analyzed header information stream.

According to the present invention, there is provided means for accumulating an input bit stream, extracting information included in a header of a layer higher than a picture layer from the input bit stream, generating header information, and converting the header information. Means for accumulating, and extracting information included in the header of the slice layer from the input bit stream, and a slice including a position where the bit stream at which the slice starts is stored and a position on the screen at which the slice starts. Means for generating information and accumulating slice information; means for adding link information indicating the position of slice information constituting a picture to picture information of header information; and starting decoding of the accumulated header information Means for pre-analyzing the header information and reading out the stored bit stream, and Means for variable length decoding the Tsu bets stream, if errors have occurred during the variable length decoding, a decoding apparatus that comprises means for performing interpolation seeking range of interpolation based on the slice information.

According to the present invention, an input bit stream is stored, information contained in a header of a picture layer or higher is extracted from the input bit stream, header information is generated, and the header information is stored. At the same time, it extracts the information contained in the header of the slice layer from the input bit stream and generates slice information including the position where the bit stream where the slice starts and the position on the screen where the slice starts Then, the slice information is accumulated, link information indicating the position of the slice information constituting the picture is added to the picture information of the header information, and the accumulated bit stream is read out before decoding starts, and the header information is read. Is analyzed in advance, the stored bit stream is read, and the read bit stream is However, when I error occurs during the variable-length decoding is a decoding method to perform interpolation seeking range of interpolation based on the slice information.

The input bit stream is stored in the buffer, and header information and slice information of each layer above the picture layer are stored in the buffer. The slice information includes the position of the start position of each slice on the screen and the address of the data on the buffer, and the information of the picture in the header information includes the slice information of the slice constituting the picture. Link information to is included. Therefore, it is possible to analyze the header information before decoding the bit stream of the MPEG2 video. Therefore, quick error processing and trick play such as fast forward and reverse reproduction can be efficiently performed.

[0026]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 shows an embodiment of the present invention.

In FIG. 1, an input terminal 1 is provided with an MPEG2
A video bit stream is provided. In this example,
The input bit stream is a Japanese digital BS
High-precision data (1920 × 108) specified in the ARIB (radio wave industry) specification for broadcasting system standards
0).

The bit stream from the input terminal 1 is supplied to a bit stream write control and header information generator 2.

The bit stream write control and header information generation unit 2 includes a CPU (Central Processing Unit) 4
Is controlled to write the input bit stream to the buffer 3 on the basis of the instruction from. With this,
Bit stream write control and header information generation unit 2
Extracts header information of a header of a picture layer or higher from an input bit stream, generates header information, and writes the header information in the buffer 3. Furthermore, the header information of the slice is extracted, and the buffer 3 storing the position of the slice on the screen and the leading data of the slice is stored.
The slice information including the upper position is generated, and the slice information is written in the buffer 3. Also, link information of information of each picture of the header information and slice information in which information of a slice included in the picture is written is added.

The header information stored in the buffer 3 is read out before starting decoding, and is supplied to the CPU 4 via the bit stream read control unit 5. The CPU 4 analyzes the header information read from the buffer 3, and decodes the header information based on the header information into the variable-length decoding unit 6, the inverse quantization unit 7, the inverse DCT transform unit 8, the motion compensation unit 9, and the like. You have made the necessary settings.

The variable length decoding section 6 is started under the control of the CPU 4. The decoding timing is set by a synchronization signal SYNC supplied from the input terminal 14 to the CPU 4. When the variable length decoding unit 6 is started, the variable length decoding unit 6 sends a data request for reading a bit stream to the bit stream read control unit 5. Upon receiving such a data request, the bit stream read control unit 5 performs a process of reading the bit stream from the buffer 3 and transferring the read bit stream to the variable length decoding unit 6.

As described above, the information of the header of the upper layer above the picture layer is stored in the buffer 3, and the information of this header is sent from the buffer 3 to the CPU 4 prior to decoding. I have. The information of the header is analyzed by the CPU 4 before decoding starts. For this reason, it is not necessary to send information of a picture layer or higher to the variable-length decoding unit 6, and the variable-length decoding unit 6 is supplied with only a bit stream of a layer lower than the slice layer.

In MPEG2, the run length of block data is variable-length coded. Further, the motion vector is variable-length coded.

The bit stream is supplied from the buffer 3 to the variable length decoding unit 6, and the data of each block and the data of the motion vector are decoded by the variable length decoding unit 6.

The decoded motion vector is stored in the error motion vector buffer 13. The error motion vector buffer 13 is used to enable high-speed interpolation processing when an error occurs.

The variable-length decoding unit 6 decodes a variable-length code and decodes a header of a layer below the slice layer.

The data of each block decoded by the variable length decoding unit 6 is inversely quantized by the inverse quantization unit 7 and
It is sent to the T converter 8. Further, the motion vector decoded by the variable length decoding unit 6 is supplied to the motion compensation unit 9.

In MPEG2, an I picture to be encoded in the same frame or field, a P picture to encode the difference between a frame or field adjacent in the forward direction, and a frame or field adjacent in both directions. A B picture for encoding the difference is sent.

When the picture to be decoded is an I picture, the image data is decoded by performing an inverse DCT transform for each (8 × 8) block by the inverse DCT transform unit 8.

If the picture to be decoded is a P picture or a B picture, the inverse DCT transform unit 8 performs an inverse DCT operation for each (8 × 8) block, and calculates a difference from the predicted image in the forward or bidirectional direction. Is decoded. The frame buffer 10 stores predicted images. The difference data decoded by the inverse DCT transformer 8 and the predicted image data motion-compensated by the motion compensator 9 based on the motion vector are synthesized, and the image data is decoded.

The decoded image data is stored again in the frame buffer 10. This frame buffer 1
0 is stored in the display processor 11.
Is sent to the display device 12.

As described above, the MPE to which the present invention is applied
In the G2 decoding circuit, the input bit stream is stored in the buffer 3, and the header information and the slice information of each layer above the picture layer are stored in the buffer 3. The slice information includes the position of the start position of each slice on the screen and the address of the data on the buffer 3, and the information of the picture in the header information includes the slice constituting the picture. Link information to the slice information is included. For this reason, MP
Prior to decoding the bit stream of the EG2 video, the header information can be read and analyzed, and quick error processing can be performed. Further, it is possible to quickly cope with the variable speed reproduction process. This will be described in detail below.

As shown in FIG. 2, the bit stream write control and header information generation unit 2
4, header information generation unit 25, slice information extraction unit 26,
And a slice information generation unit 27. The buffer 3 is divided into a bit stream area 21, a header information area 22, and a slice information area 23.

The bit stream input from the input terminal 1 is written to the bit stream area 21 of the buffer 3 and is supplied to the header information extracting section 24 and the slice information extracting section 26.

The header information extraction unit 24 extracts a header of a picture layer or higher in the bit stream.

That is, the MPEG2 stream has a hierarchical structure including a sequence layer, a GOP layer, a picture layer, a slice layer, a macroblock layer, and a block layer. Each layer is provided with a header.

The header of the sequence layer includes a start code of the sequence layer, the number of pixels in the horizontal and vertical directions of the image,
Information such as the aspect ratio is included. The header of the GOP layer includes information such as a GOP start code and time from the beginning of the sequence. The picture layer header includes information such as the picture layer start code, picture type, and quantization matrix.

The header information extraction unit 24 detects the start code of each of the layers above the picture layer, that is, the sequence layer, the GOP layer, and the picture layer. When the start code of each of these layers is detected, information included in the header is extracted. The header information of each layer of the picture layer or higher extracted by the header information extraction unit 24 is supplied to the header information generation unit 25.

The header information generator 25 generates header information as shown in FIG. 3 based on the information of the header.

That is, the sequence header includes information such as the number of pixels in the horizontal direction of the image, the number of lines in the vertical direction, and the aspect ratio. Based on this information, sequence header information 31 is generated as shown in FIG.
Further, GOP header information 32 is generated based on the information of the GOP header, and picture header information 33A and 33B are generated based on the information of the header of the picture.

As will be described later, the header information generated by the header information generation unit 25 is further added with link information indicating the address of the slice information in which the information of the slice of each picture is stored. Is stored in the header information area 22.

In FIG. 2, the bit stream from the input terminal 1 is supplied to a slice information extracting unit 26. The slice start code is detected by the slice information extraction unit 26, and the information of the slice header is extracted. This slice information is supplied to the slice information generation unit 27.

As shown in FIG. 4, the slice information generation unit 27 obtains position information on the screen at which this slice starts,
Address information of the buffer 3 in which a bit stream starting from the data of the slice is stored is stored in the slice information area 23 of the buffer 3 as slice information 35A, 35B, 35C, 35D.

That is, as shown in FIG. 5A, the slice (1) of the picture (1) starts from the position P1-1 on the screen, and the slice (2) of the picture (1) starts from the position P1-2 on the screen. Let's say it has started. Then, the data of the slice (1) of the picture (1) is stored in the address Q1 of the buffer 3, and the slice (2) of the picture (1) is stored.
Is stored at the address Q2 of the buffer 3.

As shown in FIG. 5B, suppose that slice (1) of picture (2) starts at position P2-1 and slice (2) of picture (2) starts at position P2-2. It is assumed that the data of the slice (1) of the picture (2) is stored at the address Q11 of the buffer 3 and the data of the slice (2) of the picture (2) is stored at the address Q12 of the buffer 3.

In this case, as shown in FIG. 4, in the information 35A of the slice (1) of the picture (1), P1-1 is stored as “the position of the slice (1) of the picture (1) on the screen”. Is done. Then, Q1 is stored as "the address of the slice (1) of the picture (1) on the buffer".

Information 3 of slice (2) of picture (1)
5B, P1-2 is stored as “the position of the slice (2) of the picture (1) on the screen”. And
Q2 is stored as “the address of the slice (2) of the picture (1) on the buffer”.

Information 3 of slice (1) of picture (2)
In 5C, P2-1 is stored as “the position of the slice (1) of the picture (2) on the screen”. And
Q11 is stored as the “address of the slice (1) of the picture (2) on the buffer”.

Information 3 of slice (2) of picture (2)
In 5D, P2-2 is stored as "the position of the slice (2) of the picture (2) on the screen". And
Q12 is stored as “address of slice (2) of picture (2) on buffer”.

In FIG. 2, header information (FIG. 3) composed of header information of a layer higher than the picture layer generated by the header information generation unit 25 is stored in the header information area 22 of the buffer 3.
Is written to. The slice information (FIG. 4) generated by the slice information generation unit 27 is written to the slice information area 23 of the buffer 3. Then, in the picture header information in the header information written in the header information area 22, link information indicating the address on the buffer 3 of the slice information in which the slice information of the slice included in the picture is written is added. .

For example, in the header information 33A of the picture (1) in FIG. 3, “link destination address to slice (1)”, “link destination address to slice (2)”,
Are provided, and slice information 35 of slice (1), slice (2),... Constituting picture (1) is provided here.
The addresses A1, A2 on the buffer 3 where A, 35B,... (FIG. 4) are stored are written as link information.

Similarly, the header information 33B of the picture (2) in FIG. 3 includes “link destination address to slice (1)”, “link destination address to slice (2)”,
Are provided, and slice information 35 of slice (1), slice (2),... Constituting the picture (2) is provided here.
The addresses A11 and A12 on the buffer 3 where C, 35D,... (FIG. 4) are stored are written as link information.

FIG. 6 shows the relationship among the bit stream written in the buffer 3, the header information, and the slice information.

As shown in FIG. 6, the header information area 22 stores header information of a picture layer or higher, such as a sequence header, a GOP header, and a picture header.

In the information of the picture header, an address on the buffer 3 where the slice information of the slice constituting the picture is stored as link information.

By following this link information, it is possible to reach a place in the slice information area 23 where the slice information of the slice constituting the picture is included.

The slice information stores the position of the head of the slice on the screen and the address of the buffer 3 where the data of the head of the slice is stored.

With this slice information, the position of the slice on the screen can be known, and data can be accessed from the bit stream area 21 from the beginning of the slice.

For example, when decoding the picture (1) in FIG. 5A, the picture is scanned from left to right and from top to bottom of the screen.

First, the header information 33A of picture (1) in FIG. 3 is read. The header information 33A of the picture (1) includes a “link destination address to the slice (1)”, in which the address A1 is stored.

The address A1 on the buffer 3 where the slice information of the slice (1) of the picture (1) is known from the "link destination address to the slice (1)". Thereby, slice (1) of picture (1)
Of the slice information 35A (FIG. 4).

In the slice information 35A of the slice (1) of the picture (1), P1-1 is stored as "the position of the slice (1) of the picture (1) on the screen".
Q1 is stored as “address of slice (1) of picture (1) on buffer”.

If the address Q1 of the buffer 3 is accessed in accordance with the "address of the slice (1) of the picture (1) on the buffer", data from the start position P1-1 of the slice (1) of the picture (1) is obtained. Can be read.

When the data of the slice (1) is accessed, next, the picture (1) is read from the “link destination address to the slice (2)” of the header information 33A of the picture (1) of the picture (1) in FIG. The address A2 on the buffer 3 where the slice information of the slice (2) is stored can be found.

With the address A2 of the link destination, the slice information 35B (FIG. 4) of the slice (2) of the picture (1) can be reached.

In the slice information 35B of the slice (2) of the picture (1), P1-2 is stored as "the position of the slice (2) of the picture (1) on the screen".
Q2 is stored as “the address of the slice (1) of the picture (1) on the buffer”.

If the address Q2 is accessed according to the "address of the slice (2) of the picture (1) on the buffer", data can be read from the start position P1-2 of the slice (2) of the picture (1). Can be.

Similarly, the link information in the picture header information in the header information shown in FIG.
The slice information constituting the picture can be reached, and the data can be read from the beginning of the slice based on the slice information. Thus, slices (1), (2),... Of picture (1) can be read.

In the embodiment of the present invention,
The header information and the slice information are stored in the buffer 3 together with the bit stream, and the header information is analyzed before decoding. Therefore, error processing can be performed immediately.

For example, in FIG. 5A, picture (1)
It is assumed that an error has occurred in a macroblock in slice (n) of the. In this case, the processing for the error is performed as follows.

As described above, the variable length decoding unit 6 also decodes a header below the slice layer. When finding an error, for example, by detecting a pattern that is not in the variable-length encoding table, the variable-length decoding unit 6 stops decoding at that point and reports the content of the error occurrence to the CPU 4.

The CPU 4 detects the occurrence of an error, for example, by polling the contents of a register indicating the state of the variable length decoding unit 6. Also, the CPU 4 always
The position of the macroblock being decoded by the variable length decoding unit 6 is known.

The motion vector decoded by the variable length decoding unit 6 is stored in the error motion vector buffer 13. When an error occurs, for example, the data for the previous frame is corrected using the motion vector stored in the error motion vector buffer 13, so that correction data can be created.

When the CPU 4 recognizes the occurrence of the error from the notification from the variable length decoding unit 6, when the error occurs, the CPU 4 reads out the slice header information of the next slice from the slice information area of the buffer 3, and The position information on the screen of the slice included in the information of the slice is analyzed. Then, the number of macroblocks to be interpolated is calculated from the slice position information included in the information of the next slice.

The number of macro blocks to be interpolated is, for example,
It can be obtained by the following formula.

MBNUM = (nSV-cSV) * HSIZE + (nSH-cSH) MBNUM: Number of macroblocks to be interpolated nSV: Next slice vertical position cSV: Current slice vertical position nSH: Next slice horizontal position cSH: Current slice horizontal position HSIZE: Picture horizontal size VSIZE: Picture vertical size

In the above expression, the position and size are in units of macroblocks (values obtained by dividing the position and size in pixel units by 16). If the current slice is the last slice in the picture, the next slice vertical position nSV becomes (VSIZE-1) and the horizontal position nSH becomes 0.

For example, if an error occurs at the position of slice (n) in FIG. 5A, the vertical position of the current slice is “3”, the horizontal position of the current slice (n) is “6”,
The vertical position of the next slice (n + 1) is “3”, and the horizontal position is “25”. The number of macroblocks to be interpolated in this case is given by (3-3) × the horizontal size of the picture + (25−6) = 1
9, which means that only 19 macroblocks need to be interpolated.

In this equation, the horizontal and vertical positions of the next slice are required to determine the number of macroblocks to be interpolated. The next slice position can be traced by following the link between the picture information of the header information and the slice information.

FIG. 7 is a flowchart showing the above-described error processing. In FIG. 7, first, when it is determined that an error has occurred (step S1), when the error occurs, the position of the current macroblock and the number of the picture are grasped (step S2). When the picture in which the error has occurred and the position of the macroblock are grasped, the header information of the buffer 3 is referred to, and the header information of the picture is referred to (step S3).

As shown in FIG. 3, in the header information of this picture, link information that can be traced to the slice information at the position of the macro block is written.

The current slice is what slice number,
Since the horizontal and vertical positions of the current slice are known, the slice information storing the information of the next slice is searched based on the link information of the header of this picture (step S4). As shown in FIG. 4, the horizontal and vertical positions of the next slice on the screen are known from the slice information.

When the horizontal and vertical positions of the next slice are known, the number of macroblocks to be interpolated is calculated by the above equation (step S5). Then, the variable length decoding unit 6 is set to the number of macro blocks to be interpolated, and the variable length decoding unit 6 is restarted (step S6). Then, an interpolation process is performed using the motion vector of the error motion vector buffer 13, and the process returns to the normal process.

As described above, the MPE to which the present invention is applied
In the decoding circuit of G2, the input bit stream is stored in the buffer 3, the header information of each layer above the picture layer and the slice information are stored in the buffer 3, and the slice information includes the start of each slice. The position of the position on the screen and the address of the data on the buffer 3 are included. The information of the picture in the header information includes:
Link information to the slice information of the slice constituting the picture is included. Therefore, it is possible to analyze the header information before decoding the bit stream of the MPEG2 video. Therefore, quick error processing and trick play such as fast forward and reverse reproduction can be efficiently performed.

In this example, a bit stream and
Although the header information and the slice information are all stored in the buffer 3, the memory for storing the bit stream, the header information, and the slice information may be separated.

In the above-described example, MPEG2 decoding is performed. However, the present invention can be similarly applied to MPEG1 decoding.

[0097]

According to the present invention, the input bit stream is stored in the buffer, the header information of each layer above the picture layer and the slice information are stored in the buffer, and the slice information includes The position of the start position of the slice on the screen and the address of the data in the buffer are included. The information of the picture in the header information includes the link information to the slice information of the slice constituting the picture. Have been. For this reason, MPEG2
Prior to decoding the video bitstream, it is possible to analyze the header information. Therefore, quick error processing and trick play such as fast forward and reverse reproduction can be efficiently performed.

[Brief description of the drawings]

FIG. 1 is a block diagram of one embodiment of the present invention.

FIG. 2 is a block diagram used for describing an embodiment of the present invention.

FIG. 3 is a schematic diagram used for describing header information according to the embodiment of the present invention;

FIG. 4 is a schematic diagram used for explaining slice information according to the embodiment of the present invention;

FIG. 5 is a schematic diagram used for describing the configuration of a slice according to the embodiment of the present invention;

FIG. 6 is a block diagram used for describing an embodiment of the present invention.

FIG. 7 is a flowchart used to describe an embodiment of the present invention.

FIG. 8 is a schematic diagram used to describe MPEG2.

FIG. 9 is a schematic diagram used for describing MPEG2.

[Explanation of symbols]

3 ... buffer, 4 ... CPU, 6 ... variable length decoding unit, 21 ... bit stream area, 22 ...
Header information area, 23 ... Slice information area

Continued on the front page F term (reference) 5C053 FA20 FA27 GA11 GB06 GB08 GB19 GB22 GB26 GB29 GB32 GB38 HA24 HA25 KA03 5C059 KK32 LB11 MA00 MA23 MC11 ME05 NN21 PP05 PP06 PP07 RB09 RC02 SS02 SS17 SS18 UA05 UA38 5J016 AA03 BA09 BA09 BA09 BC16 BD03

Claims (10)

[Claims] Means for accumulating an input bit stream; extracting information included in a header of a picture layer or higher from the input bit stream to generate header information; Means for accumulating header information, extracting information contained in a header of a slice layer from the input bit stream, a position where a bit stream at which the slice starts is stored, and a screen at which the slice starts Means for generating slice information including the above position and storing the slice information; and means for adding link information indicating the position of the slice information constituting the picture to the picture information of the header information; Means for reading out the stored header information before starting decoding and analyzing the header information in advance; Reading the bit stream, the decoding apparatus with the read bit stream as and means for decoding using the header information analyzed above in advance. 2. The bit stream according to claim 1, wherein the bit stream is a bit stream of MPEG1 or MPEG2 video.
3. The decoding device according to item 1. 3. Accumulating an input bit stream, extracting information included in a header of a layer higher than a picture layer from the input bit stream, generating header information, and generating the header information. Accumulating information, further extracting information included in a header of a slice layer from the input bit stream, and determining a position where a bit stream at which the slice starts is stored, and a screen at which the slice starts. The slice information including the position of the slice information is generated, the slice information is stored, and link information indicating the position of the slice information constituting the picture is added to the picture information of the header information. The header information is read out before decoding is started, the header information is analyzed in advance, and the stored bit stream is read. It reads the arm, decoding method the read bit stream so as to decode with the header information analyzed above in advance. 4. The bit stream according to claim 3, wherein the bit stream is a bit stream of MPEG1 or MPEG2 video.
Decoding method described in 1. 5. A means for storing an input bit stream, extracting information contained in a header of a picture layer or higher from the input bit stream to generate header information, and Means for accumulating, extracting the information contained in the header of the slice layer from the input bit stream, storing the bit stream where the slice starts, and the position on the screen where the slice starts Means for generating slice information including: and means for adding the link information indicating the position of the slice information constituting the picture to the picture information of the header information; and Means for reading out the header information before starting decoding and analyzing the header information in advance; Means for reading out the read stream and performing variable length decoding on the read bit stream; and means for performing interpolation by obtaining an interpolation range based on the slice information if an error has occurred during the variable length decoding. A decoding device comprising: 6. The bit stream according to claim 5, wherein the bit stream is a bit stream of MPEG1 or MPEG2 video.
3. The decoding device according to item 1. 7. The decoding device according to claim 5, wherein means for storing the variable-length-decoded motion vector is provided for the interpolation. 8. Accumulating an input bit stream, extracting information included in a header of a picture layer or higher from the input bit stream, generating header information, and accumulating the header information. And extracting information included in a header of a slice layer from the input bit stream, and a slice including a position where a bit stream at which the slice starts is stored and a position on a screen at which the slice starts. Generates information, stores the slice information, and adds link information indicating the position of the slice information constituting the picture to the picture information of the header information, and starts decoding the stored bit stream. Before reading, the header information is analyzed in advance, and the stored bit stream is read. The bit stream read by the variable length decoding, if errors have occurred during the variable length decoding, the decoding method to perform interpolation seeking range of interpolation based on the slice information. 9. The bit stream according to claim 8, wherein said bit stream is a bit stream of MPEG1 or MPEG2 video.
Decoding method described in 1. 10. The decoding method according to claim 8, wherein means for storing the variable-length-decoded motion vector is provided for the interpolation.