Image analysis – Image compression or coding – Including details of decompression
Reexamination Certificate
1998-12-07
2002-03-26
Boudreau, Leo (Department: 2621)
Image analysis
Image compression or coding
Including details of decompression
C382S232000, C382S234000, C382S246000, C382S251000, C348S014130, C348S384100, C348S403100, C358S426010, C375S240030
Reexamination Certificate
active
06363176
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to a coding and decoding system for picture data which employs orthogonal transform which is one of high efficiency compression coding methods, and more particularly to a picture data decompression apparatus which decompresses compressed picture data obtained by orthogonal transform, quantization and variable length coding of original picture data at a high speed.
2. Description of the Related Art
A procedure of processing of a conventional picture data decompression apparatus is illustrated in FIG.
12
.
Referring to
FIG. 12
, compressed picture data are first read in at step S
121
, and then at step S
122
, a header of the compressed picture data is analyzed to extract information necessary for decompression of the picture data such as a picture size, a number of blocks to be processed and so forth. Then at step S
123
, variable length codes of the compressed picture data are decoded in units of one block to obtain quantized orthogonal transform coefficients. At subsequent step S
124
, the quantized orthogonal transform coefficients are dequantized, and then at step S
125
, data obtained by such dequantization are written for one block into a memory, and the thus written data are read out from the memory. At step S
126
, the data thus read out are processed by inverse orthogonal transform to obtain picture data for one block.
The operations from the decoding processing of variable length codes at step S
123
to the inverse orthogonal transform processing at step S
126
are repeated for all of the blocks extracted in the header analysis processing at step S
122
until it is determined at step S
127
that processing for all of the blocks has been completed. Then at step S
128
, the picture data for one frame are written into a frame memory, and then the compressed picture data are decompressed. In this instance, in the decoding processing of variable length codes at step S
123
, compressed code bits are cut out by a plural number and compared with a code table, and when a coincident bit pattern is found out in the code table, coefficient data corresponding in a one-by-one corresponding relationship to the bit train in the code table can be determined. When no coincident bit pattern is found out in the code table, an additional bit is cut out, and the formerly cut out bits with the additional bit are compared with the code table again. After coefficient data are determined, they are multiplied by a quantization value for dequantization to obtain input data for inverse orthogonal transform.
Further, in the data writing and reading out processing at step S
125
, the data are written into the memory in order indicated by the numbers of the data construction in the block illustrated in
FIG. 13
, and data to be used for inverse orthogonal transform are read in order in the vertical direction and the horizontal direction of the block.
In the conventional picture data decompression method described above, a compressed code bit train is cut out by a plurality of bits, and the pattern of the thus cut out bits is compared with a code table. Then, if a coincident pattern is not found out in the code table, then the processing of cutting out a further plurality of bits from the compressed code bit train and comparing the thus cut out bits with the code table again is repeated until a coincident pattern is found out in the code table. Consequently, the conventional picture data compression method is disadvantageous in that it requires much processing time.
Further, in order to write data, which remain arranged in order as zigzag-scanned on the coding side, into a memory after dequantization, processing of scanning conversion is required, and the memory is accessed in units of one coefficient. Consequently, the conventional picture data decompression method has disadvantages also in that a very great number of processing steps are required for the microprocessor and also accessing to the memory is complicated, resulting in much processing time.
SUMMARY OF THE INVENTION
It is an object of the present invention to provide a picture data decompression apparatus which can decompress compressed data at a high speed.
In order to attain the object described above, according to the present invention, there is provided a picture data decompression apparatus for decompressing compressed picture data obtained by orthogonal transform, quantization and variable length coding of original picture data, comprising code pattern extraction means for extracting code patterns in units of n bits from the compressed picture data, n being an integral number greater than 1, m decoding and dequantization means each for decoding and dequantizing a code pattern from the code pattern extraction means, m being an integral number greater than 1, and inverse orthogonal transform means for inverse. orthogonal transforming decoded and dequantized data from the m decoding and dequantization means, each of the decoding and dequantization means decoding and dequantizing, from a code pattern of n bits extracted by the code pattern extraction means, only a predetermined variable length code of a single unique code pattern of n bits from among m different predetermined patterns.
Preferably, the code pattern extraction means includes a compressed code buffer for storing the compressed picture data, a table for defining a corresponding relationship between the m code patterns of n bits and the m decoding and dequantization means, and switching means for extracting the picture data stored in the compressed code buffer as a code pattern in units of n bits and inputting the thus extracted code pattern into a corresponding one of the decoding and decompression means referring to the table.
Preferably, the picture data decompression apparatus further comprises a dequantized data buffer for storing decoded and dequantized data from the m decoding and dequantization means, and each of the decoding and dequantization means includes a buffer address calculation section for calculating a storage address to the dequantized data buffer from a run length of a coded coefficient represented by all or some of bits of a code pattern of n bits inputted thereto, a dequantization calculation section for performing dequantization calculation of a level of a coded coefficient represented by all or some of bits of a code pattern of n bits inputted thereto referring to the dequantization table, and an extraction pointer updating section for updating an extraction pointer for the compressed code buffer by a number equal to a number of effective bits used by the buffer address calculation section and the dequantization calculation section from within a code pattern of n bits inputted thereto.
Preferably, each of the decoding and dequantization means performs storage into the dequantized data buffer in order of coded coefficients without performing scanning conversion, and the inverse orthogonal transform means reads out the stored data of the dequantized data buffer designating an address and performs inverse orthogonal transform processing of the thus read out data.
In the picture data decompression apparatus, the m different code patterns in units of n bits are decoded and dequantized simultaneously for individual code patterns by the m decoding and dequantization means, the code patterns can be decoded at a comparatively small number of steps, and consequently, the entire decompression processing can be performed at a higher speed than ever.
Further, one or a plurality of coefficients regarding data after decoding and decompression processing are stored in order of addresses into the memory without performing scanning conversion simultaneously, and upon inverse orthogonal transform, data are read in designating an address and inverse orthogonal transform of the thus read in data is performed. Consequently, decompression processing can be performed at a high speed without the necessity of processing of scanning conversion upon writing.
The above and other objects, features and
Boudreau Leo
Mariam Daniel G.
NEC Corporation
Sughrue & Mion, PLLC
LandOfFree
Picture data decompression apparatus does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Picture data decompression apparatus, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Picture data decompression apparatus will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2856531