JPH10136363A - Compressed data decoder and compressed data decoding method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To accelerate decoding by performing inverse quantization and inverse zigzag scanning transformation for only nonzero coefficients of a low frequency of generation in quantized DCT coefficients as an object and writing the result in a memory initialized to 0 beforehand. SOLUTION: An inverse quantization/inverse scanning conversion part 22 does not performs the inverse quantization and the inverse scanning conversion to a 0 coefficient outputted from a variable length decoding part 21, performs the inverse quantization and the inverse scanning conversion with only the nonzero coefficients as the object and writes the result in the memory 23. The memory 23 is entirely initialized to 0 by a memory initialization part 24 at the time of starting the decoding of a block and the inverse quantization result of the nonzero coefficients is written in the area of an address obtained by the inverse scanning conversion. When the write of the inverse quantization result is ended, an inverse DCT processing is performed to the DCT coefficient of the block, the decoding of the next block is started after the inverse DCT processing and the memory 23 is initialized to 0 again by the memory initialization part 24.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a compressed data decoding apparatus and a compressed data decoding method for decoding compressed data obtained by scan conversion, quantization, and variable length coding.

[0002]

2. Description of the Related Art In recent years, digital processing of images has become widespread. As a method for compressing and encoding digital image data, JPEG (Joint Photographic Coding Expe
rts Group) or MPEG (Motion Picture Experts g
roup).

FIG. 10 is a block diagram showing a configuration of a decoder conforming to the MPEG standard. In this MPEG decoder, MPEG encoded data input from the input terminal 1 is temporarily stored in the input buffer 2. The VLD unit 3 sequentially performs variable-length decoding on the encoded data in the input buffer 2, outputs motion vector information, quantization information, and the like therein to the inverse quantization unit 5 and the frame memory 10, and outputs the encoded data of the block. Is output to the inverse scan conversion unit 4.

The reverse scan conversion unit 4 rearranges the input one-dimensional block data into, for example, two-dimensional block data of 8 × 8 pixels with reference to the reverse scan conversion table 8. The inverse quantization unit 5 uses the quantization information input from the VLD unit 3 and the quantization table 9 to perform the inverse scan conversion unit 4.
Input from (decoded quantized DCT coefficients)
Is inversely quantized. The inversely quantized data is supplied to an inverse DCT unit 6
Are subjected to inverse DCΤ processing and output to the adder 7.

The adder 7 adds the block data of 8 × 8 pixels output from the inverse DCT section 6 to the prediction block data read from the reference image data in the frame memory 10 based on the motion vector information. Then, the result is written into the frame memory 10. The decoded images are output from the frame memory 10 in the display order.

Next, FIG. 11 shows a JPEG standard decoder.
This will be described with reference to FIG. In this JPEG decoder, JPEG encoded data input from an input terminal 11 is first subjected to variable length decoding by a VLD unit 12 and inversely quantized by an inverse quantization unit 13 with reference to a quantization table 16. The inversely quantized data is subjected to inverse scan conversion by the inverse scan conversion unit 14 based on the inverse scan conversion table 17 and rearranged into two-dimensional block data of 8 × 8 pixels.
5, after being subjected to inverse DCT processing.

In FIG. 11, the inverse scan conversion is performed after the inverse quantization, but the values in the inverse quantization table are rearranged in advance to the order after the inverse scan conversion.
After inverse scan conversion, inverse quantization may be performed.

In general, in a natural image, for example, as shown in the encoding and decoding processing examples in FIG.
Many 0 coefficients appear in the coefficients. Therefore, 2 of 8 × 8 pixels
Dimensional block data is rearranged into one dimension by zigzag scan conversion.
The variable length coding is performed as one code word by combining the coefficients.

[0009] Further, since 0 coefficients appear particularly frequently in the high-frequency component of the DCΤ coefficient after quantization, if there is no non-zero coefficient in the high-frequency component, EOB ( End of Block), and the variable length coding is terminated.

As described above, such encoded data is decoded by variable-length decoding of a codeword, inverse scan conversion, and inverse quantization. Here, the inverse scan transform and the inverse quantization are sequentially performed on each quantized DCT coefficient. For example, when one block is 8 × 8 pixels,
Inverse scan conversion and inverse quantization are performed for a total of 64 DC
Needed to be done for coefficients. Therefore, this has been a major obstacle in increasing the speed of decoding the compressed data.

[0011]

As described above, the inverse scan conversion and the inverse quantization at the time of decoding the compressed and encoded data are performed unconditionally sequentially for all DCT coefficients of one block. This has been an obstacle to speeding up the decoding of the compressed data.

The present invention has been made to solve such a problem, and it is possible to reduce the number of times of inverse scan conversion and inverse quantization when decoding compressed data, thereby realizing high-speed decoding. It is an object of the present invention to provide a compressed data decoding device and a compressed data decoding method.

[0013]

To achieve the above object, a compressed data decoding apparatus according to the present invention is characterized in that it is compressed through zigzag scanning, quantization and variable length coding. Length decoding means for decoding the compressed data in variable length, and inverse quantization for performing inverse quantization and inverse zigzag scan conversion on non-zero coefficients in the data output from the variable length decoding means in the apparatus for decoding the compressed data. A reverse scan conversion unit, a memory unit to which the output of the reverse quantization / reverse scan conversion unit is written, and a memory initialization for initializing the memory unit to 0 before the output of the reverse quantization / reverse scan conversion unit is written Means.

According to a second aspect of the present invention, there is provided a compressed data decoding apparatus for decoding compressed data in the order of zigzag scan, quantization, and variable length encoding. Variable length decoding means for performing variable length decoding, inverse quantization means for performing inverse quantization on non-zero coefficients in data output from the variable length decoding means, and non-zero coefficients which have been inversely quantized by the inverse quantization means. Reverse scan conversion means for performing reverse zigzag scan conversion; memory means for writing the output of the reverse scan conversion means; and memory initialization means for initializing the memory means to 0 before the output of the reverse scan conversion means is written. It is characterized by having.

Further, the compressed data decoding apparatus of the present invention
As described in claim 4, in an apparatus for decoding data compressed in the order of quantization, zigzag scan, and variable length encoding, variable length decoding means for variable length decoding compressed data, and variable length decoding means An inverse scan conversion unit for performing an inverse zigzag scan conversion on the non-zero coefficient in the data output from the data, an inverse quantization unit for performing an inverse quantization on the non-zero coefficient subjected to the inverse zigzag scan conversion by the inverse scan conversion unit, It is characterized by comprising memory means to which the output of the quantization means is written, and memory initialization means for initializing the memory means to 0 before the output of the inverse quantization means is written.

The DCT coefficients of the quantized block data obtained by performing variable length decoding on the compressed data are composed of a continuous number of 0 coefficients followed by non-zero coefficients. In the present invention, inverse quantization and inverse zigzag scan conversion are performed on only the non-zero coefficients among the DCT coefficients of the block data, and the result is written in a memory means which has been initialized to 0 in advance. It is possible to minimize the number of processes required for decoding data.

In order to perform inverse quantization and inverse zigzag scan conversion only on the non-zero coefficient, the address of the reference destination of the quantization table and the inverse zigzag scan conversion table for the non-zero coefficient is set, for example, by starting block decoding. It may be generated by means for counting the number of coefficients from.

As means for initializing the memory means to 0 in advance, the address information of the memory means in which data has been written is stored in the address storage means. Before data is written to the memory means, a corresponding area of the memory means may be initialized to 0 based on the address information stored in the address storage means, and the contents of the address storage means may be erased.

[0019]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of the compressed data decoding apparatus according to this embodiment.

As shown in FIG. 1, the compressed data decoding apparatus comprises a variable length decoding (VLD) 21, an inverse quantization / inverse scan converter 22, a memory 23, and a memory initialization unit 24.

The variable length decoding unit (VLD) 21 performs quantization,
Data compressed by variable-length coding and scan conversion is input, and variable-length decoding is performed on the input data to output a continuous number of 0 coefficients and a subsequent non-zero coefficient.

The inverse quantization / inverse scan conversion unit 22 does not perform inverse quantization and inverse scan conversion on the 0 coefficient output from the variable length decoding unit (VLD) 21, but performs inverse quantization only on the non-zero coefficient. And performs reverse scan conversion, and writes the result to the memory 23.

The memory 23 has been all initialized to 0 by the memory initialization unit 24 at the start of decoding a block.
The inverse quantization result of the non-zero coefficient is written in the area of the address obtained by the inverse scan conversion.

When the writing of the inverse quantization result of the non-zero coefficient of one block into the memory 23 is completed, the inverse DCT processing is performed on the DCT coefficient of the one block, and the inverse DCT processing is performed.
After the processing, decoding of the next block is started, and the memory 23 is initialized to 0 again by the memory initialization unit 24.

As described above, the compressed data decoding apparatus according to the present embodiment is configured to perform inverse quantization and inverse scan conversion only on the non-zero coefficient in the variable length decoded result of the compressed data. Therefore, it is possible to minimize the number of processes required for decoding the block data. Especially,
The number of times of processing when the data compression ratio is high and more 0 coefficients occur can be significantly reduced as compared with the conventional method.

Next, an embodiment in which the present invention is applied to a compressed data decoding apparatus of the JPEG system will be described.

As shown in FIG. 2, the compressed data decoding apparatus comprises a variable length decoding (VLD) 31 and an address generator 3
2, quantization table 33, reverse scan conversion table 3
4, a multiplier 35, a memory 36, and a memory initialization unit 37.

The variable length decoding unit (VLD) 31 is a JPEG
Input data compressed in the order of zigzag scan, quantization, and variable length coding, and perform variable length decoding,
As a result of the variable-length decoding, the number of consecutive 0 coefficients (zrn)
And the following non-zero coefficient (level) are output.

The address generation section 32 has a variable length decoding section (V
LD) Continuous number of zero coefficient output from 31 (zrn)
And the reference address of the quantization table 33 and the inverse scan conversion table 34 for the non-zero coefficient following the continuous 0 coefficient based on the variable (n) of the counter that counts the DCT coefficient that has already been inversely quantized. Generate.

The multiplier 35 has a value (quantization step size) referred to by the quantization table 33 according to the address generated by the address generator 32 and a non-zero value output from the variable length decoder (VLD) 31. Coefficient (level
1) to perform inverse quantization, and write the inverse quantization result to the memory 36 in accordance with the address obtained from the inverse scan conversion table 34 (inverse scan conversion).

Next, the details of the operation of the JPEG compressed data decoding apparatus will be described with reference to the flowchart of FIG.

First, at the start of the decoding of a block, block initialization is performed as block initialization processing.
k [] is cleared to 0 and the inversely quantized DCT
A variable n of a counter for counting coefficients from the start of block decoding is cleared to 0 (step 31).

Thereafter, the variable length decoding unit (VLD) 31 performs variable length decoding of the first code word (step 32), and determines whether or not the variable length decoding result is EOB (step 33). If the variable-length decoding result is EOB, the variable-length decoding is terminated, and the inverse DCΤ with respect to the inversely quantized DCT coefficient written in block [] of the memory 36 is obtained.
Processing is performed.

If the variable length decoding result is other than EOB, the address generation unit 32 adds zrn, which is the number of consecutive 0 coefficients output from the variable length decoding unit (VLD) 31, to the variable n of the counter (n = n). (n + zrn) (step 34), and the result of the addition is generated as a reference address of the quantization table 33 and the inverse scan conversion table 34 for the non-zero coefficient which is the n-th coefficient.

According to this address, the quantization table 33
, The quantization table value Qtable [n] for the non-zero coefficient which is the n-th coefficient is referred to, and from the inverse scan conversion table 34, the address of the memory 36 where the inverse quantization result of the non-zero coefficient is written Zigzag
[N] is obtained.

The multiplier 35 performs inverse quantization (level = level * Qtable [) by multiplying the quantization table value Qtable [n] obtained from the quantization table 33 by the value of the non-zero coefficient which is the n-th coefficient. n]) (step 35). The result of the inverse quantization is the address Zigzag [n] obtained from the inverse scan conversion table 34.
Is set to the corresponding position of block [] of the memory 36 according to (block [Zigzag [n]] = le
vel), thereby performing inverse zigzag scan conversion from one-dimensional to two-dimensional on the coefficients after inverse quantization (step 36). FIG. 4 shows the reverse scan conversion table 3
4 shows an example. 8x8 pixel two-dimensional DCT during encoding
When the zigzag scan for the coefficients is performed in the order of numbers 0 to 63 in the figure and converted into one-dimensional DCT coefficients,
The reverse scan conversion table 34 corresponds to Zigzag [] in the figure.
The content is as shown in The value on Zigzag [] is 8
The address on the × 8 two-dimensional array is shown. For example, the third coefficient (2) from the top of the one-dimensional DCT coefficient is
When the third address on Zigzag [] is referred to, it is “8”, so that the position of the second row and first column on the 8 × 8 two-dimensional array is obtained as the result of the inverse scan conversion.

Thereafter, 1 is added to the variable n of the counter (n = n + 1) (step 37). Thus, variable-length decoding, inverse quantization, and inverse scan conversion of one codeword are completed. Thereafter, the value of the variable n of the counter is checked to determine whether decoding has been completed for all coefficients of one block (step 38). When the block size is 8 × 8 pixels, the coefficient of one block is 64, and when n is smaller than 64, decoding for the next code word is repeated in the same manner. When EOB is detected or when n becomes 64 or more, the end of decoding of one block is determined, and the inverse DC $ process is started.

Conventionally, reference to the quantization table, inverse quantization operation, and reference to the inverse scan conversion table have been performed unconditionally sequentially for all DCT coefficients in the block. By performing inverse quantization and inverse scan conversion only on the coefficients, the number of processes required for decoding one block can be significantly reduced, and the decoding can be speeded up. For example, in the example shown in FIG. 12, the number of non-zero coefficients is “10”, and the number of processing is 10 /
64 will suffice.

Next, an embodiment in which the present invention is applied to an MPEG compressed data decoding apparatus will be described.
The MPEG compressed data decoding apparatus decodes data compressed in the order of quantization, zigzag scan, and variable length coding.

As shown in FIG. 5, the compressed data decoding apparatus of the MPEG system comprises a variable length decoding unit (VLD) 41, an address generation unit 42, a quantization table 43, an inverse scan conversion table 44, a multiplier 45, a memory 46, a memory initialization unit 47.

The main difference from the compressed data decoding apparatus of the JPEG system is that, as shown in FIG. 6, inverse quantization (step 66) is performed after inverse scan conversion (step 65). That is, the quantization table 43 is referred to the quantum table value according to the address referred to from the inverse scan conversion table 44 and is output to the multiplier 45. Other configurations are the same as those of the JPEG compressed data decoding apparatus shown in FIG.

By the way, in FIGS. 3 and 6, EOB is also detected at the time of decoding the first codeword.
Since the EOB code does not appear at the beginning of the block in the PEG and MPEG systems, the EOB need not be detected when decoding the first codeword.

In the JPEG system, the first code word is always the code of the DC coefficient and zrn is always 0, so that n = n in step 34 in FIG.
The processing of n + zrn can be omitted. FIG. 7 shows the procedure of the process in this case. Here, at the start of the decoding of the block, the variable n of the counter is reset to 1, and the variable length decoding, inverse quantization, and inverse scan conversion are performed on the first codeword by the processing of steps 51 to 54. From the next code word, as in FIG.
From step to step 38, variable length decoding, EOB detection, address generation (n = n + zrn), inverse quantization, inverse scan conversion, and update of counter variable n (n = n + 1) are performed.

FIGS. 8 and 9 are block diagrams showing the configuration of a compressed data decoding apparatus according to still another embodiment of the present invention.

These compressed data decoding devices are configured by adding address storage units 51 and 61 to the decoding devices shown in FIGS. 2 and 5, respectively.

The address storage units 51, 61 store information read from the inverse scan conversion tables 34, 44, that is, information on the addresses at which the inverse quantization results are written in the memories 36, 46. When the data stored in the memories 36 and 46 is used by the inverse DCT processing or the like and becomes unnecessary data, end information of the inverse DCT processing is given to the memory clear units 57 and 67.

The memory clear units 57 and 67 have the inverse DC
When the end information of the T processing is input, the address storage unit 5
By referring to the address information stored in the addresses 1 and 61, the contents of the corresponding areas of the memories 36 and 46 are initialized to 0, and the address information stored in the address storage units 51 and 61 is cleared.

In the above embodiment, the non-zero coefficient obtained by the variable length decoding of the code is multiplied by only the value of the quantization table to perform inverse quantization.
Inverse quantization may be performed by multiplying the non-zero coefficient together with quantization information added to each block, such as MPEG, in addition to the values in the quantization table.

[0050]

As described above, according to the present invention, inverse quantization and inverse zigzag scan conversion are performed only on non-zero coefficients having a low frequency of occurrence among quantized DCT coefficients obtained by variable length decoding. Is performed and the result is written in a memory initialized to 0 in advance, so that the number of times of processing required for decoding block data can be minimized, and high-speed decoding can be realized.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a compressed data decoding device according to an embodiment of the present invention.

FIG. 2 is a block diagram showing a configuration of a JPEG compressed data decoding apparatus according to the present invention.

FIG. 3 is a flowchart showing a procedure of a decoding process of the JPEG compressed data decoding apparatus of FIG. 2;

FIG. 4 is a diagram showing contents of a zigzag scan and an inverse scan conversion table at the time of encoding;

FIG. 5 is a block diagram showing a configuration of an MPEG compressed data decoding apparatus according to the present invention.

FIG. 6 is a flowchart showing a procedure of a decoding process of the MPEG-based compressed data decoding apparatus of FIG. 5;

7 is a flowchart showing another decoding processing procedure of the JPEG compressed data decoding apparatus of FIG. 2;

FIG. 8 is a block diagram showing a modification of the JPEG compression data decoding apparatus shown in FIG. 2;

9 is a block diagram showing a modified example of the MPEG compressed data decoding apparatus of FIG.

FIG. 10 is a block diagram showing a configuration of a conventional MPEG standard decoder.

And FIG. 11 is a block diagram showing a configuration of a conventional JPEG standard decoder.

FIG. 12 is a diagram illustrating a processing example of compression encoding and decoding of image data.

[Explanation of symbols]

 21 Variable length decoding unit (VLD) 22 Inverse quantization / inverse scan conversion unit 23 Memory 24 Memory initialization unit 31 Variable length decoding unit (VLD) 32 Address generation unit 33 Quantization table 34 Inverse scan conversion table 35 Multiplier 36 Memory 37 Memory initialization unit 41 Variable length decoding unit (VLD) 42 Address generation unit 43 Quantization table 44 ... Reverse scan conversion table 45 Multiplier 46 Memory 47 Memory initialization unit 51, 61 Address storage unit 57, 67 Memory clear unit

Claims (10)

[Claims] 1. An apparatus for decoding data compressed through zigzag scan, quantization, and variable length encoding, comprising: variable length decoding means for variable length decoding of the compressed data; and output from the variable length decoding means. Inverse quantization and inverse scan conversion means for performing inverse quantization and inverse zigzag scan conversion for non-zero coefficients among data; memory means for writing the output of the inverse quantization and inverse scan conversion means; and A memory initialization unit for initializing the output of the inverse quantization / inverse scan conversion unit to 0 before the output is written. 2. An apparatus for decoding data compressed in the order of zigzag scan, quantization and variable length coding, comprising: variable length decoding means for variable length decoding of the compressed data; and output from the variable length decoding means. Inverse quantization means for performing inverse quantization on the non-zero coefficient of the data, inverse scan conversion means for performing an inverse zigzag scan conversion on the non-zero coefficient inversely quantized by the inverse quantization means, A compressed data decoding apparatus comprising: a memory unit to which an output of a unit is written; and a memory initialization unit that initializes the memory unit to 0 before an output of the inverse scan conversion unit is written. 3. An apparatus for decoding data compressed in the order of zigzag scan, quantization, and variable length encoding, comprising: variable length decoding means for variable length decoding the compressed data; and output from the variable length decoding means. Inverse quantization means for performing inverse quantization on the non-zero coefficient of the data, inverse scan conversion means for performing an inverse zigzag scan conversion on the non-zero coefficient inversely quantized by the inverse quantization means, Memory means to which the output of the means is written; address storage means for storing the address information of the memory means to which the output of the inverse scan conversion means is written; and before the output of the inverse scan conversion means is written to the memory means. The corresponding area of the memory means is set to 0 based on the address information stored in the address storage means.
And a means for erasing the contents of the address storage means. 4. An apparatus for decoding data compressed in the order of quantization, zigzag scan, and variable-length coding, wherein said variable-length decoding means outputs variable-length decoding means for decoding said compressed data. Inverse scan conversion means for performing an inverse zigzag scan conversion on the non-zero coefficient of the data obtained, inverse quantization means for performing an inverse quantization on the non-zero coefficient subjected to the inverse zigzag scan conversion by the inverse scan conversion means, A compressed data decoding apparatus comprising: a memory unit to which an output of a quantization unit is written; and a memory initialization unit that initializes the memory unit to 0 before an output of the inverse quantization unit is written. 5. An apparatus for decoding data compressed in the order of quantization, zigzag scan, and variable length coding, comprising: variable length decoding means for variable length decoding the compressed data; and output from the variable length decoding means. Inverse scan conversion means for performing an inverse zigzag scan conversion on the non-zero coefficient of the data obtained, inverse quantization means for performing an inverse quantization on the non-zero coefficient subjected to the inverse zigzag scan conversion by the inverse scan conversion means, Memory means to which the output of the dequantizing means is written; address storage means for storing the address information of the memory means to which the output of the dequantizing means is written; and the output of the dequantizing means being written to the memory means. Before, based on the address information stored in the address storage means, a corresponding area of the memory means is initialized to 0. Means for erasing the contents of the address storage means. 6. A method of decoding data compressed through zigzag scan, quantization, and variable length coding, wherein the compressed data is variable length decoded, and the inverse of the variable length decoded data to a non-zero coefficient. A compressed data decoding method characterized by performing quantization and inverse zigzag scan conversion, and writing the result into a memory initialized to 0 in advance. 7. A method for decoding data compressed in the order of zigzag scan, quantization, and variable length encoding, wherein the compressed data is variable length decoded, and a non-zero coefficient of the variable length decoded data is inverted. After quantization, inverse zigzag scan conversion is performed on the inversely quantized non-zero coefficient,
A compressed data decoding method, wherein the result is written in a memory initialized to 0 in advance. 8. A method for decoding data compressed in the order of zigzag scan, quantization, and variable length encoding, wherein the compressed data is variable length decoded, and a non-zero coefficient of the variable length decoded data is inverted. After quantization, inverse zigzag scan conversion is performed on the inversely quantized non-zero coefficient,
The result is written in a memory initialized to 0 in advance, and the information of the write address is stored in an address storage unit, and when the memory is initialized, the information is written based on the address information stored in the address storage unit. A compressed data decoding method, wherein a corresponding area of a memory is initialized to 0 and the contents of the address storage unit are deleted. 9. A method for decoding data compressed in the order of quantization, zigzag scan, and variable-length encoding, wherein the compressed data is variable-length decoded, and the inverse of non-zero coefficients in the variable-length decoded data is determined. A compressed data decoding method, comprising: performing a zigzag scan conversion, dequantizing the non-zero coefficient subjected to the inverse zigzag scan conversion, and writing a result of the inverse zigzag scan conversion to a memory initialized to zero in advance. 10. A method for decoding data compressed in the order of quantization, zigzag scan, and variable length coding, wherein the compressed data is variable length decoded, and the inverse of non-zero coefficients of the variable length decoded data is obtained. After performing the zigzag scan conversion, the non-zero coefficient subjected to the inverse zigzag scan conversion is inversely quantized, and the result is written in a memory initialized to 0 in advance, and information of the write address is stored in an address storage unit. Wherein when the memory is initialized, a corresponding area of the memory is initialized to 0 based on the address information stored in the address storage unit, and the contents of the address storage unit are erased. Data decryption method.