JP3045696B2 - Decoding method and apparatus - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a decoding method and an apparatus used for transmitting information via a transmission line or storing the information in a storage device.

[0002]

2. Description of the Related Art In recent years, with the progress of the information-oriented society, there has been an increasing demand to transmit moving images to others over time and distance barriers. With the era of full-scale commercialization of digital technology, it has become possible to record and reproduce moving images with a recording device and to transmit data over long distances using a communication network. In addition, not only in the field of communication but also in the field of broadcasting, transmission and coding schemes using digital technology have been adopted.

[0003] Since the moving picture and audio signals of digital signals generally have a large code amount, it is essential to use a high-efficiency coding technique for efficient recording and transmission. The device has already been prototyped.

A video CD in which digital moving images are recorded on a CD (compact disk), and a DVD in which digital moving images of higher quality and longer time are recorded than video CDs
Is one of the application examples. These videos C
In a decoding device for playing back D or DVD, special playback such as fast forward and rewind playback is essential, and in order to realize these, an international standard called Information Technology (MPEG) is called "Information Technology". Coding of Moving Pictures
And Associated Audio for Digital Storage Media at Up to About 1.5Mbits "" InformationTechnology-C
oding of moving pictures and associated audio for
digitalstorage media at up to about 1,5Mbits / s (IS
O / IEC11172-2) was used in some cases. Hereinafter, a conventional digital moving picture decoding method using the above-described method will be described with reference to the drawings.

[0005] First, an encoding method and a bit stream of a digital moving image in MPEG will be described.
In the MPEG, it is assumed that a digital moving image is composed of a series of video frames 700, and as shown in FIG.
Encode 0. The sequence is called a group of pictures (hereinafter, referred to as a GOP), and is usually encoded by being divided into a series of video frames 600 every about 0.5 seconds.

[0006] As an example, as schematically shown in FIG. 5B, a GOP is an I picture (Intra-Picture: an intra-coded image) encoded using data only in the video frame. The P picture (P picture) that can be predictively coded from the data of the previous picture (I picture or P picture)
(redictive Picture: inter-frame forward prediction coded image)
B pictures (Bidirectionally) that can be interpolatively predicted encoded from temporally preceding and succeeding I and P pictures.
Predictive-Picture: an inter-frame bidirectional predictive coded image).

FIG. 6A shows the structure in each picture. A picture is made up of a continuous area in a band shape on a screen called one or more slices, and a slice is made up of an image block 800 of one or more 16-pixel by 16-line pixels called one or more macroblocks. . As shown in FIGS. 6 (B) and 6 (C), the macro block is composed of a plurality of image blocks called 8 blocks of 8 horizontal pixels × 8 vertical lines. As an example, four luminance signals and two color signals ( Cb,
Cr) Each consists of one block.

As described above, the bit stream has a hierarchical structure, and the sequence, GOP, picture, and slice, which are the code strings constituting the higher hierarchy, in the hierarchical structure, are respectively represented on the bit stream. A uniquely identifiable start code is provided, and an area for holding header information for holding the coded information of the code string constituting each layer and for holding layer information called an extension is provided. As an example, these are arranged as shown in FIG.

In the hierarchy below the macro block, a macro block address increment indicating how far the macro block is from the position of the previously encoded macro block, and a macro block indicating the prediction mode information of the macro block selected at the time of encoding. A type, a quantizer scale indicating a quantization width, a motion vector used for motion compensation, a coded block pattern indicating which block is coded and present in a bit stream, a coded DCT (Discrete Cosine) Transform: Discrete cosine transform) Information such as coefficient information exists.

A variable length code is used for encoding each information below the hierarchy constituted by these macroblocks, and the higher the frequency of occurrence, the shorter the code is assigned, so that the majority of the bit stream is occupied. Hierarchical information constituted by macroblocks is efficiently encoded.

In practice, a bit stream is divided into packets 40 each having an appropriate length as shown in FIG. Separately, the encoded audio signal is similarly packetized, and the packetized information is provided with a packet header 41 having identification information at the beginning of each packet so that they can be distinguished from each other. The packets are multiplexed to form a bit stream 1000. Of this packet 40,
The area containing the information originally contained in the bit stream 1000 is called the payload area 42.

Next, a method for fast-forwarding and rewinding the bit stream will be described. In the case of normal playback, all the above bit streams are played, all pictures are decoded, and images are displayed.
At the time of fast-forward playback, all bit streams are transferred to the decoding device, and I pictures among them are selected and decoded and displayed by the decoding device, or only I pictures are selectively sent to the decoding device. The I picture is supplied and decoded by the decoding device and displayed.

The decoding apparatus uses a code table in which the correspondence between codes and the corresponding symbols of image data and the like is previously shown, and decodes the codes by collating the code strings sequentially input to this table. It is going to be.

In this decoding device, when a symbol that is not originally defined in the code table is found, decoding of the picture is immediately stopped, and a process called an error concealment process is performed on the picture as an error process.

As an example of the error concealment processing, for a picture for which decoding has been interrupted, for an incomplete image portion, the contents of a picture which has already been decoded are copied and supplemented to complete a video frame, and decoding is interrupted. It shall be displayed in place of the picture that has been set. Then, processing for skipping to the header of the next picture is performed.

Alternatively, as another example of the error concealment process, it is assumed that the picture whose decoding has been interrupted is not displayed at all, and the process of simply skipping to the header of the next picture is performed.

In the above two error concealment processes, one of the processes may always be performed.
Further, the processing may be adaptively switched.

Similarly, if the presence of the header and extension does not conform to a predetermined grammatical rule, it is regarded that an error has occurred, and the same processing as described above is performed.

However, in practice, when the above method is used, the analysis capability of the bit stream is insufficient, and it is complicated to select an I-picture. Therefore, as shown in FIG. Instead of supplying the entire bit stream fragment including the entire bit stream 1000 to the decoding device,
As shown in FIG. 9 (B), only bit stream fragments 300, 310, 320, etc. composed of packets, which are considered to include a predetermined code string such as an I picture, are selectively supplied to the decoding device. As shown in (C), the payload areas of the packets constituting this bit stream fragment are rearranged to form an elementary bit stream 1001, and only this I picture is played back in chronological order one after another. Often do. This reproduction is performed by sequentially collating the codes constituting the elementary bit stream 1001 with a code table, and decoding the code string into data for reproduction. In the code table,
The correspondence between the code used for encoding and the data for reproduction is defined. In some cases, a P picture may be reproduced in addition to an I picture of the elementary bitstream 1001.

The rewind reproduction is realized by transferring the supplied discrete bit stream fragments back in time. That is, for example, the bit stream fragments 320, 310, 300,... Are supplied to the decoding device in this order. At this time, the packet including the head of the I picture included in the GOP can often be identified by using management information separately recorded on a disk or the like. In particular, data of an I picture is often packetized so that the head of the I picture becomes the head of the packet.

[0021]

However, in the above configuration, at the time of fast-forward or rewind reproduction, discrete bit stream fragments are supplied to the decoding device in packet units. FIG. 9 is an enlarged view of a portion indicated by.
As shown in (D), when the bit stream fragment does not completely include the code sequence of the picture, the code is originally decoded in the area before and after the connection point A with the subsequent bit stream fragment. Will be different from the code to be converted. At this time, if the codes in the area before and after the connection point A are grammatically correct, that is, codes that can be decoded by the code table, the connection point A is normally detected as an error during decoding. Despite the fact that it is necessary to end decoding of the picture that has been decoded up to this point, perform some error processing on this image, and start decoding anew on the picture that is placed after that. Since no error can be detected, the area before and after the connection point A is also determined to be a normal code and is decoded, and the subsequent picture is decoded as a code different from the original code. In some cases, subsequent pictures are not decoded normally, and an image different from the original picture is decoded. As a result, significant image distortion occurs.

Further, in this case, error detection is not always performed before and after the point A of the connection part, and the elementary bit stream part where a grammatically clear error occurs is analyzed and decoded. However, there has been a problem that the decoding device wastes time and cannot secure a time necessary for a normal decoding operation thereafter.

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems, and provides a decoding method and apparatus capable of performing a normal decoding operation at the time of special reproduction such as fast-forward or rewind. The purpose is to:

[0024]

According to the decoding method of the present invention, a plurality of first code strings obtained by dividing a main code string having a hierarchical structure are inputted, and the plurality of first code strings are inputted. In addition to arranging the columns sequentially, a part of the code sequence before the first layer information associated with the predetermined code sequence among the code sequences forming the predetermined hierarchy in the main code sequence. And a step of forming a third code string by arranging a predetermined second code string such that at least one of the codes including the character string does not correspond to a predetermined code table used for decoding the code string. Things.

Further, after the step of forming the third code string, the third code string is decoded using the predetermined code table and includes a part of the second code string. When the code does not correspond to the predetermined code table, a step of setting the code string as an error code is provided.

When a code including a part of the second code string is set as an error code, the decoding process is stopped, and the second code including a part of the error code is included. The decoding process is performed again from the first layer information accompanied by the code string.

Further, a predetermined error processing is performed on a code string that has been decoded immediately before the code string used as the error code.

Further, the predetermined code sequence is predetermined picture information, the first hierarchical information is header information, and the plurality of input first code sequences are arranged in time sequence. The main code string including the plurality of pieces of predetermined picture information obtained above is divided into a plurality of pieces and includes the predetermined picture information.

Further, the predetermined picture information is any one of I picture information, I and P picture information, or I, P and B picture information.

Further, the second code string is inserted between the first hierarchical information and information immediately before the first hierarchical information.

Further, the second code string is replaced with a code string located immediately before the first hierarchical information.

Further, the second code string is composed of continuous 0 data of 23 bits or more.

Also, the decoding apparatus according to the present invention receives a plurality of code strings each including a first code string obtained by dividing a main code string having a hierarchical structure, and inputs the plurality of code strings. Extracting means for extracting and outputting a first code string from each of them, and a starting position of hierarchical information associated with a predetermined code string among code strings constituting a predetermined hierarchy included in the output of the extracting means And a second code string generating means for generating a second code string such that one of the codes including a part thereof does not correspond to a plurality of predetermined code tables used for decoding the code string. Selecting means for selecting and outputting the output of the extracting means and the output of the second code string generating means; and immediately before the start position of the hierarchical information detected by the detecting means, the second code string Control the selecting means so as to arrange It is obtained so as to comprise a control means. Further, the second code string is composed of continuous 0 data of 23 bits or more.

[0034]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiment. FIG. 1 is a diagram for explaining a decoding method according to an embodiment of the present invention. The operation of the decoding method of the present invention will be described below. First, at the time of fast-forward or rewind playback, not all bit stream fragments including the entire bit stream 1000, which is the main code string, are supplied to the decoding device, but as shown in FIG.
The discrete bit stream fragments 300, 31 which are considered to include a given picture, here in particular an I picture
0, 320, etc. are sequentially supplied to the decoding device, and the I pictures included in the discrete bit stream fragments are successively reproduced.

Similarly, the rewind reproduction is to transfer and arrange the supplied discrete bit stream fragments so as to precede in time, form an elementary bit stream, and sequentially reproduce the elementary bit stream. Is realized. That is, as an example, the bit stream fragments 320, 31
Are supplied to the decoding device in the order of 0, 300,.

The bit stream fragment input to the decoding device is composed of one or a plurality of packets 40 including the entire I picture or a part of the I picture.
In the packet 40, a first code string obtained by dividing a bit stream 1000 as a main code string is stored in a payload area 41, and a packet header is attached to each of the payload areas 41. Then, as shown in FIG. 1B, only the code string stored in the payload area 41 is extracted from each packet 40 as a first code string, and the extracted code string is sequentially traced back in time series or in time series. As shown in FIG. 1 (C),
A bit stream of a digital moving image used for fast-forward playback or rewind playback, that is, an elementary bit stream 1001 as a third code string is obtained.

FIG. 2 is a diagram showing main steps of the decoding method according to the embodiment of the present invention. Payload area 41
When arranging the code string stored in the bit stream 1000, as shown in FIG. 2, a bit stream fragment is analyzed, and a predetermined picture, Then, the dummy data 10 is inserted as a second code string immediately before the header attached to the I picture. The dummy data 10 is, for example, a code formed by connecting a plurality of 0 data. By adding the dummy data 10, as shown in FIG. 2, a plurality of 0 data continue between the header of the I picture and the immediately preceding elementary bit stream.

The dummy data 10 will be described. FIG.
0 is a diagram for explaining the principle of the decoding method according to the embodiment of the present invention, and FIGS. 10 (A) to 10 (E) show different joining patterns between bit stream fragments, respectively. 10 (F) indicates a code table. Here, a case is shown where six types of codes correspond to six symbols. In the figure, 22a to 22a
d is the last code of the bit stream fragment 20; 23 is the picture header of the I picture of the bit stream fragment 21 arranged after the bit stream fragment 20; Is provided. Here, the case where the first two bits of at least 0 are used as the picture header is particularly described. FIG. 10 (A) to (D)
In this example, dummy data 10a consisting of 10-bit 0 is arranged immediately before this header. Here, for the sake of explanation of the principle, dummy data 10a composed of 10-bit 0 is used as an example, but dummy data composed of different codes is actually used.

Here, in particular, bit stream fragment 2
0 is assumed to be cut off in the middle of the data of the I picture,
It is assumed that the bit stream fragment 21 has an I picture header 23 and an I picture sequentially arranged from the top. Although the case where the bit stream fragment 20 is cut off in the middle of the data of the I picture is described, the same principle can be applied to the case where the bit stream fragment 20 is cut off at the B picture or P picture.

FIG. 10 illustrates a case where an I picture header is arranged immediately after a bit stream fragment 20, but the same applies to a case where a picture header is arranged in the middle of a bit stream fragment. Explainable.

First, after the code up to immediately before the last code string of the bit stream fragment 20 is decoded, the last code is decoded. Here, considering the case where no dummy data is provided, as shown in FIG. 10 (E), when this last code 22a is "011", when compared with the code table, the code consisting of "011" is obtained. Is not in the code table shown in FIG. 10 (F) by itself,
By combining with "0" at the leading end of the picture header 23, it becomes "0110" and is decoded as a symbol E. As a result, the leading end of the picture header 23 disappears, and the picture header 23 is not recognized as a normal picture header in subsequent decoding processing. When the code obtained by combining the rest of the picture header 23 and the head of the subsequent I picture matches the code in the code table, these are sequentially decoded, and the I picture is also decoded normally. May not be performed.

As shown in FIG. 10A, the last code 2
Even when 2a is "011", when this code 22a is compared with the code table, the code consisting of "011" is not alone in the code table shown in FIG. Is combined with "0" of the symbol E to be decoded to the symbol E as "0110". However, the remaining leading portion of the dummy data 10a has a code “0000” that does not exist in the code table.
, It is determined as an error code in normal decoding processing.

In the case shown in FIG. 10B, the last code 22b of the bit stream fragment 20 is "01".
In the case of, the code consisting of “01” is not in the code table by itself when compared with the code table, and the code “0100” that does not exist in the code table even when combined with “00” at the beginning of the dummy data 10a , It is determined as an error code in normal decoding processing.

As shown in FIG. 10C, the last code 22c of the bit stream fragment 20 is "0111".
In the case of, when this is compared with the code table, this “0111”
Is independently decoded as a symbol P. However, since the subsequent dummy data 10a is composed of a code that does not exist in the code table, it is determined to be an error code in normal decoding processing.

As shown in FIG. 10 (D), when the last code 22d of the bit stream fragment 20 is "0", when this code is collated with the code table, the code consisting of "0" alone appears in the code table. No, and dummy data 10
Such a code does not exist in the code table even when it is combined with “000” at the leading end of “a”. Therefore,
A code including the last code 22d and the first portion of the dummy data 10a is determined as an error code in normal decoding processing.

The I picture header has a code "00000000" called a start code, which can be prefixed with zero or more bits "0" called "stuffing bits".
0000 0000 0000 0001 ”, so that a search for a start code that satisfies this condition after being determined to be an error code in each of the above cases ensures that the subsequent I-picture header can be reliably captured. It is possible.

Therefore, even when the code string itself is not in the code table and its head part is decoded in combination with the immediately preceding code string, the head part is always stored in the immediately preceding code string. By arranging, as dummy data, a code string in which a part combined with or one of the remaining parts is not in the code table, one of the codes including the dummy data can be detected as an error code. it can.

As a result, the decoding is temporarily stopped at the position where the dummy data is provided, that is, at least immediately before the picture header, and the decoding can be performed again from the subsequent I picture header. It is possible to perform normal decoding processing without loss of pictures.

Here, when a normal decoding device finds an error code not in the code table as described above, it immediately stops decoding the picture and performs error concealment processing. As an example of the error concealment processing, for a picture whose decoding has been interrupted, for an incomplete image portion, the contents of the already decoded picture are copied and supplemented to complete a video frame, and decoding is interrupted. It shall be displayed instead of a picture. Then, processing for skipping to the header of the next picture is performed.

Alternatively, as another example of the error concealment processing, it is assumed that the picture whose decoding has been interrupted is not displayed at all, and the processing of simply skipping to the header of the next picture is performed.

The above two error concealment processes may be performed by either one of them at all times.
The processing may be switched adaptively. Therefore, by arranging such dummy data, detection of an error code is guaranteed at the position where the dummy data is arranged, and abnormal decoding is performed in the vicinity of the connection between bit stream fragments. Even if it does not analyze the bitstream portion that is grammatically erroneously located after this connection portion, the decoding device does not waste time and can perform normal decoding operation thereafter. The required time can be secured.

In particular, in the MPEG standard and the like, FIG.
As in the example shown in FIG. 0, it is possible to bring a picture header at the beginning of a bit stream fragment. Therefore, an error can be generated by inserting dummy data into a connection portion of a bit stream fragment in which an abnormality is likely to occur. Therefore, especially in the MPEG standard, since the decoding device can reliably notify the discontinuous position of the connection part of the bit stream fragment, an image different from the original image is not decoded, and the occurrence of a significant image disorder is prevented. Can be prevented.

Also, in this case, it is ensured that error detection is performed at the connection part, so that it is not necessary to analyze even a bitstream part which is located after the connection part and has a grammatically clear error. In addition, the decoding device does not waste time, and the time required for the subsequent normal decoding operation can be secured.

Next, dummy data used when MPEG2, which is a higher standard of MPEG, is used as an encoding method will be described. FIG. 11 shows a code table of variable length codes for representing DCT coefficient information used in the MPEG2 standard.
The DCT coefficient information indicates the number of consecutive 0s and the magnitude of non-zero coefficients that appear after zigzag scanning of the coefficients of a certain image block, respectively, by a run.
, Level.

Combinations of runs and levels with a high appearance frequency are shown in the code table shown in FIG. 11A, and combinations with a low appearance frequency are shown with a run and a level following an escape code. 11 (B), FIG. 11 (C)
, Respectively. FIG.
In (A), the last s of the code represents the level code,
It is assumed that 0 is positive and 1 is negative. That is, (run, level) = (0, −3) corresponds to the code “0010 11”, and (run, level) = (2, −2046) corresponds to the code “0”.
000 01 0000 10 1000 0000
0010 ".

That is, for the DCT coefficient information, the maximum code length of 0 data that can be obtained is (run, level) =
“0000 01 0000 0” corresponding to (0,1)
0 0000 0000 0001 "is 17 bits, and 0 data longer than this does not appear. In other words, regarding DCT coefficient information, continuous 0 data of 18 bits or more is an error code, and No matter how it is mixed in the code word corresponding to the symbol, an error code is generated.Since there is no continuous 0 data of 18 bits or more in the code table, the code itself is an error code.

In MPEG, since a start code consisting of 23 bits of 0 data and 1 bit of 1 data is arranged at the head of the header of a picture, the start code described with reference to FIG. Similarly to the state, even if the last bit of the bit stream fragment and the first 17-bit 0 data of the dummy data are recognized as a certain code word, if the dummy data is 18-bit 0 data, Since 0 bits are continuous for 24 bits between the remaining 1-bit 0 data of the dummy data and the start code in the picture header, and such a code word having 24 bits of continuous 0 data does not exist in the code table, this part is Is determined as an error code, and a new picture decoding process starts from the start code of the picture header in which the dummy data is located immediately before. But so that the start.

That is, from this, by placing continuous 0 data of 18 bits or more as dummy data immediately before the header of a predetermined picture layer, the immediately preceding bit stream fragment is cut by DCT coefficient information. In such a case, the error code can be reliably detected.

Here, the description has been made assuming that the bit stream fragment is composed of DCT coefficients. However, actually, in the MPEG2 standard, there is a possibility that up to 22 pieces of 0 data may continue in a portion called an extension. Therefore, by inserting 0 data of 23 bits or more, for example, 32 bits of 0 data or the like as dummy data at the time of decoding, the decoding device can reliably detect data including dummy data as an error code.

Also, in MPEG, variable-length codes are generally used for information below the macroblock layer, including the DCT coefficients, and continuous 0 data of 23 bits or more is defined as being prohibited from use except for the start code. ing. From this, if 0 data of 23 bits or more is used as the dummy data, it is possible to make the decoding device discover the dummy data itself as an undefined symbol and detect an abnormality.

FIG. 3 shows a block diagram of a decoding apparatus used in the decoding method according to the present embodiment. In FIG. 3, an input terminal 100 to which a code string is input is connected to a formatter 110, a buffer 120 is connected to receive an output of the formatter 110 as an input, and
Numeral 30 is connected so that the output of the buffer 120 becomes an input, and the output of the decoding means 130 is connected to the decoding output terminal 140.

The operation of the decoding device configured as described above will be described below. First, the input terminal 100
The digital moving image input as the bit stream fragment packetized in the above is input to the formatting means 110, where the format processing is performed.

FIG. 4 shows the formatting means 110. 4, reference numeral 200 denotes an input terminal, 205 denotes an extraction unit, 2
10 is a header detection means, 220 is a dummy data generation means, 230 is a control means, 240 is a selection means, and 250 is an output terminal.

In the formatting means 110, the packetized code string input to the input terminal 200 is input to the extraction means 205, and only the payload portion of the packet is extracted by the extraction means 205, and the header detection means 210 and the selection means Output to 240. A header position of a header of a predetermined picture, for example, an I picture, of the output is detected by the header detection unit 210, and the header position is notified from the header detection unit 210 to the control unit 230, and the dummy data generation unit 220 It generates dummy data and instructs the selection means 240 to output it. Here, the header detecting means 210 does not analyze the picture data at all, but performs only a simple I-picture header detecting operation.

The control means 230 receives the output from the header detection means 210 and instructs the selection means 240 to insert the dummy data immediately before the header of the I picture.
Output of extraction means 205 and dummy data generation means 220
Control to select the output of And selecting means 2
The output of the extracting means 205 and the output of the dummy data generating means 220 selected at 40 are output from the output terminal 250.

The output of the formatting means 110 is temporarily buffered in the buffer 120 and then output to the decoding means 1.
The digital video signal is input to the decoding unit 130, the digital video is decoded in the decoding unit 130, and output from the output terminal 140.

As the decoding means 130, a general decoding means for performing decoding using a code table is used. FIG. 12 shows an example of the decoding means 130. Here, the decoding means usually used for MPEG is described. The code string input to the input terminal 131 is buffered once in the buffer 132 and input to the code string decoding unit 133. The code string decoding means 133 decodes and outputs image information such as motion vectors and DCT coefficients. At the time of fast-forward or rewind reproduction, the code string decoding means 133 selectively decodes only I picture data. This decoding is performed by using a code table to convert a code string into a corresponding symbol, for example, image data or the like. The detection of the error code formed by arranging the dummy data is performed by the code string decoding unit 133. The decoded motion vector is supplied to the motion compensation unit 134. The decoded image information is supplied to the inverse quantization means 135.

The image information is inversely quantized by the inverse quantization means 135, and orthogonal transform decoding such as inverse DCT is performed by the orthogonal transform decoding means 136. The result and the image data in the image memory 137 for storing the decoded image data are added in the adder 138 and stored in the image memory 137. Here, the decoded image data is data obtained by motion-compensating the decoded data stored in the image memory 137 using the motion vector in the motion compensation unit 134.

The decoded image data is read from the image memory 137 in the order in which the decoded image is to be output, and output from the decoding output terminal 139. Here, when the code string decoding means 133 of the decoding means 130 finds a symbol which is not originally defined, the decoding of the picture is immediately stopped and the processing of skipping to the header of the next picture is performed.

At the same time, when the decoding is stopped,
An error concealment process is performed. Error information is output from the code string decoding unit 133 to the control unit 129,
In response to this information, an error concealment instruction signal is output from the control unit 129 to the motion compensation unit 134, and the motion compensation unit 134 performs error concealment processing.

In the error concealment process, the motion compensating means 134 uses the image memory 137 for the picture for which decoding has been interrupted and for the unfinished image portion.
, The contents of the already decoded picture are copied and supplemented to complete the video frame, and stored in the image memory 137 in place of the picture whose decoding has been interrupted.

Alternatively, as another example of the error concealment processing, it is assumed that the picture whose decoding has been interrupted is not stored in the image memory 137 at all, and the partially stored image portion of the picture whose decoding has been interrupted is invalidated. Free image memory. One of the two error concealment processes may always be performed, and a configuration in which the process is adaptively switched is also possible.

As described above, the error concealment processing and the processing of skipping to the picture header are performed on the input data. Also, when the presence of the header or extension does not conform to the grammar rules, the same processing is performed. Therefore, an error code formed by adding dummy data as in the present embodiment can be detected by such ordinary decoding means.

As described above, according to the present embodiment, one of the codes including a part of the code string immediately before the picture header accompanying the I picture corresponds to the predetermined code table used for decoding the code string. Dummy data 10 that no longer works
Is inserted, an error code can be generated immediately before the picture header by inserting the dummy data 10, and the decoding unit 130 can reliably start decoding for each I picture. Even when the vicinity of the connection part between the bit stream fragments is recognized as a code different from the original code and decoded, the I picture is not decoded into an image different from the original code, and significant image disturbance occurs. It can be prevented before it happens.

In this case, it is guaranteed that error detection is performed immediately before the I picture header.
Since it is not necessary to analyze even a bitstream part where a grammatically clear error occurs after the connection part between the bitstream fragments, the decoding means 130
However, no time is wasted, and the time necessary for the subsequent normal decoding operation can be secured.

Here, a modification of the present embodiment will be described with reference to FIG. 13, the same reference numerals as those in FIG. 13 indicate the same or corresponding parts. In the above embodiment, after extracting only the data necessary for decoding stored in the payload area of the packet constituting the bit stream fragment, dummy data is inserted immediately before the header of the picture. As shown in FIG. 13, the data immediately before the picture layer is replaced with dummy data 10.

The decoding method according to this modification is similar to the decoding method shown in FIG.
In the formatting means 110 shown in FIG.
30 can be performed using a decoding device that controls the selection means 240 to replace the data immediately before the picture with dummy data.

Even in such a modification, as described above, an error is generated by replacing the dummy data,
As in the above-described embodiment, the decoding device can reliably restart the decoding process from the picture following the dummy data, and the image different from the original image is not decoded.
Significant image disturbance can be prevented from occurring.

Further, since the data immediately before the picture header is replaced with dummy data, the number of bits of the elementary bit stream is not increased by the dummy data, and the data transfer load is reduced as compared with the above embodiment. be able to.

Although the above embodiment has been described with reference to the case where decoding is performed using the decoding apparatus as shown in FIG. 3, the present invention has the same means as the above-described decoding apparatus. The decoding may be performed using a computer or the like programmed as described above. In such a case, the same effect as in the above-described embodiment can be obtained.

In the above embodiment, the case where dummy data consisting of continuous 0 data is used has been described. However, in the present invention, the dummy data is at least a code including a part or all of the code string. As long as one of the code strings satisfies the condition that the code table does not correspond to the code table used for decoding, another code string such as continuous one data may be used. It has the same effect as.

However, one of the advantages of using continuous 0 data as dummy data in the above embodiment is that when a continuous normal bit stream is input during normal reproduction, the start of the picture header 0 data is inserted in front of the code. Normally, when some 0 data is arranged immediately before the header, the start code in the header is regarded as so-called stuffing data and ignored. For this reason, the dummy data consisting of 0 data is regarded as stuffing data and is skipped. As a result, when decoding the bit stream, it is not recognized as an error code, error processing is not performed during normal reproduction, and error processing is not performed. The burden is reduced.

In the above embodiment, the case where discrete bit stream fragments are supplied to the decoding device at the time of fast forward or rewind reproduction has been described. However, the present invention is not limited to this. The present invention is also applicable to a case where an original continuous bit stream fragment which is not discrete is supplied.

Further, in the above-described embodiment, the case where only an I picture is decoded and displayed as a predetermined code string at the time of fast forward or rewind reproduction has been described, but the present invention is not limited to this. Instead, it is also possible to decode and display I-pictures and P-pictures, and to display all types of pictures. For example, when decoding and displaying all pictures, By inserting or replacing the dummy data immediately before the picture header, the same effect as in the above embodiment can be obtained.

In the above embodiment, the insertion and replacement of the dummy data are performed at the position immediately before the picture header accompanying the I picture in the hierarchy constituted by the pictures of the bit stream. The present invention may be arranged such that dummy data is arranged before layer information such as a header and an extension attached to a predetermined code string of a code string constituting a predetermined layer of a bit stream. Dummy data may be arranged before the picture header. In this case, the header detecting unit 210 detects the positions of all the picture headers. A similar effect is achieved.

Further, the decoding method and apparatus according to the present embodiment employs “Information Technology Generic Coding of Moving Pictures and・ Associated Audio for Digital Information "" I
nformation technology-Generic coding of moving
pictures and associated audio for digital informat
ion "(ISO / IEC13818-2), the decoding method is also called" video codec for audiovisual services, commonly called H.261. "
At px 64kb / s "" Video Codec for Audiovisual se
rvices at px 64kb / s "(CCITT Recommendation H.261)
Can also be applied to the decoding method employing.

Further, in the above embodiment, the case of decoding and displaying a digital moving image has been described. However, the present invention is not limited to this, and decoding and reproduction of digital audio using a code table, The present invention can also be applied to decoding of a data signal using another variable length code, and in such a case, the same effects as those of the above-described embodiment can be obtained.

[0088]

As described above, according to the present invention, a predetermined hierarchy is formed in the main code string of the first code string obtained by dividing the main code string having a hierarchical structure. Immediately before the first layer information associated with a predetermined code string of the code strings, one of codes including a part of the code string is:
By inserting a second code string that does not correspond to a predetermined code table used for decoding a code string, an error code is generated immediately before the first layer information by inserting the second code string. As a result, the decoding can be reliably started for each predetermined code string, so that the first code string is supplied at the time of fast-forward or rewind reproduction, and the vicinity of the connection portion between the first code strings is provided. Even if is recognized as a code different from the original code and is decoded, the predetermined code sequence is not decoded into data different from the original code, and normal decoding is performed during special playback such as fast forward or rewind. There is an effect that a decoding method and apparatus capable of performing a simple decoding operation can be provided.

Further, this ensures that error detection is performed immediately before the first hierarchical information, so that the grammatical position located after the connection between the first code strings of the third code string. Since it is no longer necessary to analyze the part where an error is apparently generated, the decoding means does not waste time, and the time required for the subsequent normal decoding operation can be secured.

Further, since the second code string is replaced with a code string located immediately before the first layer information of the third code string, the number of bits in the third code string is There is an effect that the load at the time of data transfer can be reduced as compared with the above embodiment without being increased by the code string of 2.

The predetermined code sequence is predetermined picture information, the first layer information is header information, and the plurality of input first code sequences are a plurality of first code sequences arranged in time series. Since the predetermined picture information is included in the payload area of the packet obtained by dividing the main code string including the predetermined picture information into a plurality of pieces, the vicinity of the connection portion between the payloads is the original code. Even when the decoding is performed by recognizing a code different from that of the original image, the predetermined picture is not decoded into an image different from the original image, so that an effect of remarkably disturbing the image can be prevented.

[Brief description of the drawings]

FIG. 1 is a diagram for explaining a decoding method according to an embodiment of the present invention.

FIG. 2 is a diagram for explaining main steps of a decoding method according to the embodiment of the present invention.

FIG. 3 is a block diagram of a decoding device according to the embodiment of the present invention.

FIG. 4 is a block diagram of a format unit of the decoding device according to the embodiment of the present invention.

FIG. 5 is a diagram for explaining a conventional encoding method.

FIG. 6 is a diagram for explaining a conventional encoding method.

FIG. 7 is a diagram for explaining a conventional encoding method.

FIG. 8 is a diagram for explaining a conventional encoding method.

FIG. 9 is a diagram for explaining a conventional decoding method at the time of fast forward and fast rewind reproduction.

FIG. 10 is a diagram for explaining the principle of the decoding method according to the embodiment of the present invention.

FIG. 11 is a diagram showing a code table for describing a decoding method according to the embodiment of the present invention.

FIG. 12 is a block diagram for explaining a decoding unit of the decoding device according to the embodiment of the present invention.

FIG. 13 is a diagram for describing a modification of the decoding method according to the embodiment of the present invention.

[Explanation of symbols]

 10, 10a Dummy data 20, 21 Bit stream fragment 22a to 22d Last code 23 I picture header 40 Packet 41 Payload area 42 Packet header 80 Bit stream fragment connection area 100, 200 Input terminal 110 Formatting means 120 Buffer 130 Decoding Conversion means 131 input terminal 132 buffer 133 code string decoding means 134 motion compensation means 135 inverse quantization means 136 orthogonal transform decoding means, 137 image memory 138 adder 139 output terminals 140, 250 output terminal 205 extraction means 210 header detection means 220 Dummy data generation means 230 Control means 240 Selection means 300, 310, 320 Bit stream fragments 500 Sequence 600 Group of pictures 700 Video frame Frame 800 macroblock 900 block 1000 bitstream 1001 elementary bitstream

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int.Cl. ⁷ , DB name) H04N 5/91-5/956 G11B 20/18 G11B 20/10-20/12 H03M 7/30 H03M 13 / 00 H04N 7/24-7/68

Claims (11)

(57) [Claims] 1. A plurality of first code strings obtained by dividing a main code string having a hierarchical structure are input, and the plurality of first code strings are sequentially arranged. At least one of the codes including a part of the code string before the first layer information associated with the predetermined code string among the code strings forming the predetermined layer in
A decoding method, comprising: arranging a predetermined second code string so as not to correspond to a predetermined code table used for decoding a code string to form a third code string. 2. The decoding method according to claim 1, wherein after the step of forming the third code sequence, the third code sequence is decoded using the predetermined code table. A decoding method, comprising: when a code including a part of the second code string does not correspond to the predetermined code table, sets the code string as an error code. 3. The decoding method according to claim 1, wherein when a code including a part of the second code string is set as an error code, the decoding process is stopped and the code set as the error code is used. Decoding processing is performed again from the first layer information accompanied by the second code string, a part of which is included. 4. The decoding method according to claim 1, wherein a predetermined error process is performed on a code string that has been decoded immediately before the code string used as the error code. . 5. The decoding method according to claim 1, wherein the predetermined code string is predetermined picture information; the first hierarchical information is header information; Of the main code string divided into a plurality of pieces including the plurality of pieces of the predetermined picture information arranged in time series, and including the predetermined picture information. Characteristic decoding method. 6. The decoding method according to claim 1, wherein the predetermined picture information is I picture information, I and P picture information.
A decoding method characterized by being picture information, or one of I, P and B picture information. 7. The decoding method according to claim 1, wherein the second code string is inserted between the first layer information and information immediately before the first layer information. 8. The decoding method according to claim 1, wherein said second code string is replaced with a code string located immediately before said first hierarchical information. 9. The decoding method according to claim 1, wherein the second code string is continuous 0 data of 23 bits or more. 10. A plurality of code strings each including a first code string obtained by dividing a main code string having a hierarchical structure, and a first code string is input from each of the plurality of code strings. Extracting means for extracting and outputting a code string; detecting means for detecting a start position of hierarchical information associated with a predetermined code string among code strings constituting a predetermined layer included in an output of the extracting means; A second code string generating means for generating a second code string such that at least one of the codes including a part does not correspond to a plurality of predetermined code tables used for decoding the code string; Selecting means for selecting and outputting the output of the second code string generating means, and immediately before the start position of the hierarchical information detected by the detecting means,
A decoding unit for controlling the selection unit so as to arrange the second code string. 11. The decoding device according to claim 10, wherein the second code string is continuous 0 data of 23 bits or more.