JPH1023424A - Device for decoding compression data of moving image - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce the disturbance of a decoded display picture as small as possible when a decoding error is generated in a variable length code. SOLUTION: At the time of decoding a variable length code from the code bit string of moving image compression data, a variable code decoding part refers to a variable length code table, and when a corresponding variable length code is not entered, generates a decoding error detection signal. When an Nth picture in decoding is an intra-frame encoded picture data (I picture) or inter- frame forward direction predictive encoded picture data (P picture), a movement compensating part substitutes unit data corresponding to the (N-1) th I or P picture stored in a memory area for a unit data area (slice line data) including an error occurance block to obtain a decoded display picture. When the picture in decoding is a bidirectional predictive encoded picture data (B picture), unit data corresponding to the (N-1) th picture movement-compensated and stored in a display memory are substituted for the unit data including the error occurance block, so that the display of the slice line data including the error occurance block and having the possibility of incorrect data can be suppressed.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a decoding device for compressed image data, and more particularly to a decoding device for decoding a compressed moving image data.

[0002]

2. Description of the Related Art In the JPEG, MPEG1, and MPEG2 standards known as image compression coding systems, still image data or moving image data subjected to discrete cosine transform (DCT) is quantized, and the obtained quantized DCT is obtained.
The coefficients are further subjected to variable length coding using Huffman coding or the like to obtain compressed data.

When a decoding device for decoding the image data compressed in this way is constituted by an LSI, the decoding LSI comprises a variable-length code decoding unit for decoding the variable-length code of the compressed image data, And an inverse discrete cosine transform unit for performing inverse discrete cosine transform on the inversely quantized DCT coefficients to obtain image data.
In the case of MPEG1 and MPEG2, since a moving image is compressed, a motion compensation unit for performing motion compensation on image data obtained by inverse discrete cosine transform is further provided.

The variable-length code decoding section of the decoding LSI has a variable-length code table such as a Huffman code as shown in FIG. 4, and refers to this table to convert a variable-length code bit string into a variable-length code. And the corresponding run level output, that is, decoded data is obtained. In addition,
Of the run level outputs, the level
Is a DCT coefficient that is not “0” in the code bit string obtained by zigzag scanning the quantized DCT coefficient, and “run” represents the number of “0” s preceding the level. I have.

When the variable length code decoding section refers to the variable length code table shown in FIG. 4 and a variable length code corresponding to a code bit string is not entered, a variable length code decoding error occurs. Becomes Then, when such a decoding error occurs, the subsequently decoded run level output lacks reliability because the cutout position of the variable length code and the like are incorrect.

Therefore, generally, in a decoding device,
For example, IEICE Technical Report ICD94-86 (1994-0)
8) As shown in “Development of LSI for decoding variable-length code compatible with 110 MHz MPEG2”, processing for searching for the next header (slice header in the case of MPEG) is performed. Then, based on the next header obtained by the search, a correct decoding process for the variable-length code bit string after this header is restored.

[0007]

However, even when a bit error or the like occurs in the code bit string of the compressed image data or when the variable length code is not correctly cut out by the variable length code decoding unit, the actual In many cases, a variable length code in the variable length code table hits these bit strings. When the code in the variable-length code table hits in this way, an incorrect run-level output is continued because a variable-length code decoding error is not detected. Then, when such an illegal variable-length code hit continues and eventually no longer hits, a decoding error is detected for the first time.

Therefore, the reproduced image data obtained by decoding during the period from when an error actually occurs before the detection of a decoding error to when a decoding error is detected and a normal run level output is obtained is incorrect. Becomes

Further, in the image compression data decoding apparatus, it is necessary to decode the image compression data obtained from the reception or the storage medium and supply the decoded display images to the display unit sequentially for display. Is required to be performed more quickly. Therefore, conventionally, in a device for decoding image compressed data, each circuit block is pipelined,
The data after the decoding process in each circuit block is sequentially passed to the subsequent circuit block to finally obtain a decoded reproduced image.

Therefore, when a decoding error is detected at the time of decoding a certain block in the variable-length code decoding unit, for example, the motion compensation unit uses the data with a high probability of being incorrect several blocks before and performs motion compensation using these data as a reference image. May have done so. In this way, by performing decoding processing on incorrect data, not only wasteful processing is increased, but also the disorder of the decoded image due to a decoding error is larger, and the disorder continues for a longer period of time. There is also the possibility that the quality will be reduced.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a compressed image decoding apparatus capable of solving the above-mentioned problem and making the disturbance of a decoded image as inconspicuous as possible at the time of a decoding error of a variable length code.

[0012]

Means for Solving the Problems In order to achieve the above object, a moving picture compression data decoding apparatus according to the present invention has the following features.

That is, when a decoding error of a variable-length code is detected in a variable-length code decoding process on a code bit string of compressed video data of a variable-length code, a predetermined error to which decoding error-occurring data belongs is detected. Is replaced with the data of the corresponding unit data area of the past decoded and reproduced image, and a decoded and reproduced image is formed based on the data created by the replacement.

In a specific configuration, an I-frame for storing a reference I-picture obtained by decoding an intra-frame coded image is stored.
It has a picture memory, a P picture memory for storing a reference P picture obtained by decoding an inter-frame forward coded image, and a display memory for storing decoded display image data to be displayed.

In the above configuration, if the image being decoded at the time of occurrence of a decoding error is an intra-frame coded image or an inter-frame forward coded image, a unit data area (for example, an error-occurring block) to which decoding error-occurring data belongs. For example, one slice line of the MPEG standard)
Replace with the data of the corresponding unit data area of either the past reference I picture or reference P picture stored in the picture or P picture memory.

On the other hand, if the image being decoded is a bidirectionally predictive coded image in which motion compensation is performed based on a reference I picture and a reference P picture, the unit data area to which the block in which a decoding error has occurred belongs is displayed in the display area. Replace with the data of the unit data area corresponding to the past decoded display image data stored in the memory.

In the above configuration, the I-picture memory, the P-picture memory, and the display memory are each provided with a data holding unit for holding data for a unit data area. If no decoding error is detected during the decoding of the unit data area, the data of the held unit data area is overwritten to the corresponding memory, and if a decoding error occurs during the decoding of the unit data area, Discards the held data of the unit data area and maintains the data of the past corresponding unit data area stored in each memory.

[0018]

Embodiments of the present invention (hereinafter, referred to as embodiments) will be described below with reference to the drawings.
Note that the same reference numerals are given to portions corresponding to the drawings described above, and description thereof will be omitted.

[Structure of Decoding Apparatus for Compressed Video Data] FIG. 1 shows a schematic configuration of a decoding apparatus for compressed video data according to the present embodiment. In this embodiment,
The decoding device is pipelined.

The code bit string of the compressed moving image data obtained by storage media or communication is used for header detection /
The header of each layer is supplied to the parameter separation unit 10, and first, a header of each layer is detected from a code bit string having a hierarchical structure. Here, the hierarchical structure of the compressed image data is, for example, MPEG1
Takes a six-layer structure as shown in FIG.
The uppermost sequence layer has an SH (Sequence Header) and a GOP (Group of Pictures). The GOP layer has an I picture (intra-coded image data) and a B picture (bidirectional prediction) which are the minimum units of a screen group. Encoded picture data) and P pictures (inter-frame forward-predicted encoded picture data). And
Each picture layer is divided into slice lines (unit data areas) having a minimum length of 16 pixels and having an arbitrary length as a unit having a start code, and this slice layer further includes an arbitrary number of macro blocks (MB). And the macroblock layer is composed of six blocks of 8 × 8 pixels.

The header detecting / parameter separating unit 10 shown in FIG. 1 separates a code bit string into parameters based on the detected header of each layer, and sets a block of 8 × 8 pixels as a minimum decoding processing unit for each block. Are output to the variable-length code decoding processing unit 12.

The variable-length code decoding processing unit 12 cuts out the variable-length code from the supplied variable-length code bit string with reference to the variable-length code table shown in FIG.
A level output, that is, a quantized DCT coefficient code string is output. If the corresponding variable length code is not entered with reference to the variable length code table, the variable length code decoding processing unit 12 outputs a variable length code decoding error detection signal (for example, a Huffman error signal as shown in FIG. 1). A detection signal is generated and output to the motion compensation unit 18 at the subsequent stage.

The quantization D from the variable length code decoding processor 12
The code sequence of the CT coefficients is supplied to the inverse quantization unit 14 where it is inversely quantized, and the inversely quantized DCT coefficient is inversely transformed by the inverse discrete cosine transform unit 16 into data for each pixel before DCT transformation. You.

If the image data obtained by the inverse discrete cosine transform is an I-picture (intra-frame coded image), the motion compensator 18 outputs this to the display memory 26 of the image memory 20 as it is. Is also output to the memory area 22 for use. The I picture memory area 22 stores the data of the I picture as the reference image 1 until the next I picture is obtained.

In the case of a P picture (inter-frame forward prediction coded image), the motion compensator 18 outputs P picture data to the P picture memory area 24 and reads I picture data from the memory area 22. Motion compensation is performed by adding to the P picture data, and the obtained decoded display image is temporarily stored in the display memory 26. The P picture data output to the P picture memory area 24 is stored as the reference picture 2 until the next P picture is obtained, like the I picture.

When the image data obtained by the inverse discrete cosine transform is a B picture (bidirectional predictive coded image), the I and P picture data stored in the memory areas 22 and 24 are converted to B picture data. The added and bidirectional motion compensation is performed, and the obtained decoded display image is temporarily stored in the display memory 26.

As described above, I picture, P picture, B picture
The decoded display image after motion compensation sequentially obtained based on the picture is displayed on a display (not shown), for example, as [I picture, B picture, B picture, P picture, B picture, B picture, P picture... I picture]. They are displayed in the order they correspond.

[Compensation for Occurrence of Decoding Error of Variable Length Code] Next, a method of compensating for the occurrence of a decoding error of a variable length code, which is a feature of the present embodiment, will be described. FIG. 3 shows this error compensation method. As shown in the figure, the feature of the present embodiment is that when a decoding error is detected at the time of decoding a variable length code, the data of the entire slice line, which is a unit data area to which the decoding error belongs, is stored in the image memory 20. Is to replace the data of the corresponding slice line of the past picture.

More specifically, the variable length code decoding processing section 12 shown in FIG. 1 refers to the variable length code table and detects no variable length code decoding error if there is no entry corresponding to the variable length code bit string. A signal is generated, and this error detection signal is output to the motion compensator 18. When an error flag is entered in the variable-length code table of FIG. 4, when a decoding error of the variable-length code is detected, this error flag is added, and this And a motion compensation unit 18 via an inverse discrete cosine transform unit 16
May be supplied.

When the error detection signal is directly supplied from the variable length code decoding processing unit 12 or the data having the error flag is supplied from the inverse discrete cosine transform unit 16, the motion compensation unit 18 has an error occurrence block. The slice line to which the error-occurring block belongs is replaced with the slice line of the immediately preceding picture stored in the image memory 20 according to the type of the picture.

(In the case of an I or P picture) First, if the picture being decoded by the variable length code is an I picture or a P picture, and a decoding error detection signal is The unit 18 processes the block before the error occurrence block. Also, I
Memory areas 22, 24 for pictures or P pictures
Stores the (N-1) -th I or P picture data immediately before the N-th picture being decoded.

Therefore, the motion compensator 18 detects the slice line (i-th) to which the error-occurring block belongs based on the decoding error detection signal from the variable-length code decoder 12, and determines whether the slice line belongs to an I picture or a P picture. The data of the corresponding (i-th) slice line is read out from the memory areas 22 and 24 of the memory block and replaced with the data of the i-th slice line of the error-occurring block. The slice line to which the error-occurring block belongs and the slice line of the corresponding past picture can be easily determined by detecting the header added to the head thereof.

As for the I picture, image data obtained by replacing the slice line to which the error-occurring block belongs with the slice line data of the immediately preceding I picture is stored in the display memory 26 as a decoded display image.
It is stored in the memory area 22 as the N-th I picture.

In the case of a P picture, the motion compensator 18
Stores the new P picture obtained by the replacement in the memory area 2
4 and store P picture data in the memory area 22.
The motion compensation is performed by adding the I-picture data read out from the memory and stored in the display memory 26 as a decoded display image.

Next, the memory area 22,
In a configuration in which the decoded image data is stored in 24,
When a decoding error detection signal is generated, the beginning of the data of the slice line to which the block in which the error occurred belongs has already been overwritten in each of the memory areas 22 and 24, and the immediately preceding (N-1) th data may have been lost. . Therefore, in the case of such a configuration, it is preferable to provide a save area such as a buffer as a data holding unit that temporarily stores data before storing data for one slice line in the memory areas 22 and 24. If the decoding error of the variable length code is not detected during the decoding of the data of one slice line sequentially stored in the save area, the slice line data is moved to each of the memory areas 22 and 24 and the corresponding slice line data is moved. Update the slice line data of the past picture. Conversely, if a decoding error is detected during decoding of the data for one slice line, the data stored in the save area is discarded, and the past picture stored in the memory areas 22 and 24 is deleted. The corresponding slice line data is maintained without being rewritten.

When the correlation between the I picture and the P picture is strong, for example, when a variable length code decoding error occurs during the decoding of the Nth I picture, the Nth I picture May be replaced with the corresponding slice line data of the (N-1) -th P picture which is closer to and closer to the image data. Similarly, when the error-occurring picture is the N-th P picture, if the I-picture is more similar to the (N-1) -th P picture, the picture is replaced with the corresponding slice line data of the I-picture. Good.

(In the case of a B picture) Next, when the decoded data is a B picture, the B picture is not a reference picture and is not stored in any of the memory areas 22 and 24 as described above. Therefore, when the picture being decoded is a B picture, the motion compensation unit 18 generates a decoding error detection signal of a variable length code in a predetermined block, and the display memory 2
6, the corresponding slice line data is read from the immediately preceding motion-compensated decoded display image data stored in
The slice line data to which the error-occurring block belongs is replaced, and this is stored in the display memory 26 as a decoded display image.

Further, in the case of a configuration in which the data is decoded and stored in the display memory 26,
A save area for temporarily storing data for a slice line is provided. If no decoding error of the variable length code is detected for the data of one slice line once stored in the save area, the slice line data is moved to the display memory 26. If a decoding error is detected, the data in the save area is discarded, and the corresponding past slice line data stored in the display memory 26 is maintained as it is.

[0039]

According to the structure of the present invention, when a decoding error of a variable length code is detected, the data of the unit data area including the occurrence area, for example, all of the slice line data are transferred to the past normal picture. By replacing the data with the data in the corresponding unit data area, it is possible to avoid the problem that the decoded display image is affected for a long period of time due to the occurrence of a decoding error, or large image disturbance is caused.

In addition, when an error is detected, the data is replaced with the corresponding unit data of a past picture that is updated relatively frequently without using data of another area in the same picture. Also in this case, compensation is performed so that the error area becomes a natural image.

Therefore, in the present invention, the disturbance of the decoded display image due to the variable-length code decoding error can be made as inconspicuous as possible, and the display quality can be easily maintained.

[Brief description of the drawings]

FIG. 1 is a diagram showing a schematic configuration of a moving image compression data decoding device according to an embodiment of the present invention.

FIG. 2 is a schematic diagram for explaining a hierarchical structure of image data in the MPEG system.

FIG. 3 is a diagram for explaining an error compensation method when a decoding error occurs according to the embodiment of the present invention.

FIG. 4 is a diagram showing a configuration of a variable length code table.

[Explanation of symbols]

Reference Signs List 10 header detection / parameter separation unit, 12 variable length code decoding processing unit, 14 inverse quantization unit, 16 inverse discrete cosine transform unit, 18 motion compensation unit, 20 image memory, 22
Memory area for I picture, memory area for 24P picture, 26 display memory.

Claims (5)

[Claims] 1. A decoding apparatus for decoding a compressed moving image data to obtain a decoded reproduced image, wherein a decoding error of a variable length code is detected in a variable length code decoding process on a code bit string of the compressed moving image data composed of a variable length code. When detected, the data of the predetermined unit data area to which the decoding error occurrence data belongs is replaced with the data of the corresponding unit data area of the past decoded and reproduced image, and A compressed moving picture data decoding apparatus for forming a decoded and reproduced image based on the data. 2. The compressed video data decoding apparatus according to claim 1, wherein an I picture memory for storing a reference I picture obtained by decoding the intra-frame encoded image, and an inter-frame forward encoded image. Reference P obtained by decoding
A P-picture memory for storing pictures, and a display memory for storing decoded display image data to be displayed, wherein the image being decoded when the decoding error is detected is an intra-coded image or an inter-frame forward coding. In the case of an image, the data of the unit data area to which the decoding error occurrence data belongs is stored in a corresponding unit of either a past reference I picture or a reference P picture stored in the I picture or P picture memory. A moving image compression data decoding apparatus, wherein the data is replaced with data in a data area. 3. The moving picture compression data decoding device according to claim 1, wherein an I picture memory for storing a reference I picture obtained by decoding an intra-frame coded image, Reference P obtained by decoding forward encoded image
A P-picture memory for storing a picture, and a display memory for storing decoded display image data to be displayed, wherein an image being decoded when the decoding error is detected moves based on the reference I picture and the reference P picture. In the case of a bidirectionally predicted coded image for which compensation is performed, the data of the unit data area to which the decoding error-occurring block belongs is replaced with the data of the unit data area corresponding to the past decoded display image data stored in the display memory. A compressed moving picture data decoding apparatus characterized in that: 4. The moving picture compression data decoding device according to claim 2, wherein the I picture memory, the P picture memory, and the display memory hold data of a unit data area, respectively. If the decoding error is not detected during the decoding of the unit data area, the data of the unit data area held in the data holding unit is overwritten to the corresponding memory, and the unit data is written. When the decoding error occurs during the decoding of the area, the data of the unit data area held in the data holding unit is discarded, and the data of the past corresponding unit data area stored in each memory is deleted. A moving image compressed data decoding apparatus characterized by maintaining. 5. The moving picture compression data decoding apparatus according to claim 1, wherein said unit data area is one slice line of the MPEG standard. apparatus.