JPH0470060A - Method and device for restoring picture data - Google Patents

Abstract

PURPOSE:To restore the pictures at a high speed in hierarchical restorage by inhibiting the adverse quantization, the adverse DCT conversion, and the updating of a picture memory when no effective coefficient is included in a block. CONSTITUTION:The input code data are decoded by a decoding means 1 and turned into a quantization coefficiency. If no significant coefficient is included in the decoced quantization coefficient within a single block, the zero data are outputted as the restoration data. If the quantization coefficient contains a significant coefficient, an adverse quantization means 2 and an adverse DCT conversion means 3 are actuated for restoration of pictures. Then the restored picture data are selected by a selection means 6 and stored in a picture memory 7 at a 1st stage. At and after a 2nd stage, the picture data stored in the memory 7 are added to the picture date restored by the means 3 and this addition output is selected by the means 6 and stored in the memory 7. Meanwhile the block addresses are counted up and no addition is carried out in a block when the significant coefficient is not contained in the quantization coefficient within a block. Thus the pictures are restored at a high speed.

Description

[Detailed Description of the Invention] [Summary] A signal based on an orthogonal transform encoding method for a multi-valued image in which a multi-valued image is divided into blocks each consisting of a plurality of pixels, the pixels within the block are orthogonally transformed, and then encoded. Regarding an image data restoration device that restores images, the purpose of the present invention is to provide an image data restoration device that can reduce the number of circuits and increase speed.
quantizing transform coefficients obtained by performing two-dimensional discrete cosine transform on the tone values of the plurality of N×N pixels for each block;
A device for restoring an image from encoded data obtained by encoding the obtained quantized coefficients includes a decoding means for decoding the inputted encoded data, and a dequantizing means for dequantizing the quantized coefficients decoded by the decoding means to obtain DCT coefficients. and inverse DCT of the DCT coefficients inversely quantized by the inverse quantization means.
When there is no significant coefficient among the quantized coefficients in the block decoded by the inverse DCT transform means to transform and the decoding means,
The apparatus is configured to include significant coefficient detection means for controlling output as restored image data without performing inverse quantization by the inverse quantization means and inverse DCT transformation by the inverse DCT transformation means.

[Industrial Application Field] The present invention relates to a device for restoring a data-compressed image, and more specifically, the present invention relates to a device for restoring a data-compressed image, and more specifically, it divides a multivalued image into blocks each consisting of a plurality of pixels, and after orthogonally transforming the pixels in the block, The present invention relates to an image data restoring device for restoring a signal based on an orthogonal transform coding method for a multivalued image to be coded.

[Prior Art] As a highly efficient compression method for image data, there is, for example, an adaptive M-loss cosine transform encoding method. Adaptive Discrete Cosine Transform Coding
Cosine Transform: ADCT
), for example, is obtained by dividing an image into blocks of 8 x 8 pixels, converting the image signal of each block into coefficients of a spatial frequency distribution by two-dimensional discrete cosine transformation, and quantizing it with a darkness value adapted to visual perception. This is a method of encoding using a Huffman table in which quantization coefficients are statistically determined.

FIG. 6 is a block diagram of an ADCT encoding circuit. The encoding operation will be explained in detail below using FIG.

The image is divided into blocks each consisting of 8×8 pixels as shown in the original image signal table of FIG. The two-dimensional DCT transform unit 32 orthogonally transforms the input image signal by DCT transform, and converts the input image signal into the eleventh
It is converted into coefficients of the spatial frequency distribution shown in the DCT coefficient chart in the figure. Then, it is output to the linear quantization section 33.

FIG. 7 is a block diagram of the two-dimensional DCT transform section. The input image signal is first converted by a one-dimensional DCT conversion section 32-1. Then, the transposing unit 32-2 transposes the rows and columns, and the one-dimensional DCT transforming unit 32-3 transforms the rows and columns again.
These two one-dimensional DCTs perform row and column transposition in 2-4.
A two-dimensional DCT transformation is performed. Note that the one-dimensional DCT conversion unit 32-1, 32-3 performs the conversion using conversion constants.

The linear quantization unit 33 converts the input DCT coefficients into the 12th
Quantized dark value holding unit (quantization matrix) 34 configured with the dark values shown in the quantized dark values for general CT coefficients in the figure.
Linearly quantize by dividing by the value of .

FIG. 13 is a chart of DCT coefficients (quantized coefficients) after quantization, which is the result of quantization using the constants shown in FIG. 12 as an example of dark values regarding DCT coefficients. As shown in FIG. 13, DCT coefficients below the dark value become 0, and quantized coefficients are generated in which only the DC component and a few AC components have values.

The two-dimensionally arranged quantized coefficients are converted into one-dimensional one by zigzag scanning. FIG. 14 is an explanatory diagram of the scanning order (zigzag scan) of the quantized coefficients mentioned above. 1
When the block is 8×8, for example, dots are selected in sequence in the upper left, then to the right, then to the lower left, and so on, diagonally to the lower left.

The data selectively and sequentially read out by the zigzag scan is input to the variable length encoding unit 35.
variable-length encodes the difference between the DC component at the beginning of each block and the DC component of the previous block.

For the AC component, the values (indexes) of effective coefficients (coefficients whose value is not 0) and the run length (run) of invalid coefficients (coefficients whose value is 0) up to that point are variable-length coded for each block. Each of the DC and AC components is encoded using a code table 36 consisting of a Huffman table created based on statistics for each image, and the obtained code data is sequentially output from a terminal 37.

On the other hand, the code data described above is restored to an image by the following method. FIG. 8 is a block diagram of an ADCT type restoration circuit. Code data input from the terminal 40 is input to the variable length decoding section 41 . The variable-length decoding unit 41 decodes the input code data into fixed-length data of index and run using a decoding table 42 that is an inverse table to the Huffman table of the code table 36 described above.
Output to 3. The inverse quantization unit 43 multiplies the input quantization coefficient by the value stored in the inverse quantization matrix 44 to obtain D.
The CT coefficients are restored and output to the two-dimensional inverse DCT transform unit 45. The two-dimensional inverse DCT transform unit 45 orthogonally transforms the input DCT coefficients by inverse DCT transform, and transforms the coefficients of the spatial frequency distribution into an image signal.

More specifically, the two-dimensional inverse DCT transform unit 45 will be explained.

FIG. 9 is a block diagram of a two-dimensional inverse DCT transform section. The DCT coefficients input from the terminal 40 are subjected to one-dimensional inverse DCT transformation in a one-dimensional inverse DCT transformation section 41-1, and output to a transposition section 41-2. The transposing unit 41-2 transposes the rows and columns of the coefficients within one block and outputs the result to the one-dimensional inverse DCT transform unit 41-3.

The one-dimensional inverse DCT transform unit 41-3 again performs one-dimensional inverse DCT transform on the input transposed coefficients, and outputs the result to the transpose unit 41-4. The transposing unit 41-4 again performs 1 in the same way as the transposing unit 41-2.
Swaps the rows and columns of coefficients within a block. The signal obtained by the above operation is output from the terminal 46. That is, the image is restored. Note that the aforementioned one-dimensional inverse DCT transform units 41-1 and 41-3 perform the transform using inverse DCT transform constants.

The above-mentioned restoration will be explained in more detail.

FIG. 15 is a block diagram of a conventional restoration circuit (hierarchical restoration). A conventional hierarchy restoration method will be explained using FIG. 15. The encoded data is input from a terminal 51 to a variable length decoding section 52 . The scanning order of DCT coefficients is XOI, XO2,...
X64 (corresponding to 1 to 64 in FIG. 14), the variable length decoding unit 52 first calculates the first term to the n1 term (nl
) quantized code data corresponding to DCT coefficients D
The CT coefficients are decoded using the decoding table 53. The decoded quantized coefficients are input to the dequantization section 54, and the dequantization section 54 dequantizes them into DCT coefficients using a quantization matrix 55. The inversely quantized DCT coefficients are converted into image data by a two-dimensional inverse DCT transformer 56. The obtained image data is held in an image memory 58 via an adder 57, and is sent to a terminal 5.
Output from 9. Next, the variable length decoding unit 52 decodes the code data corresponding to the DCT coefficients from the n1th term to the n2th term (nl<n2≦64) into quantized DCT coefficients.

The decoded quantized coefficients are output to the inverse quantization section 54 and D
Inverse quantization into CT coefficients. The inversely quantized DCT coefficients are converted into image data by a two-dimensional inverse DCT transformer 56.

Then, the adder 57 adds the obtained image data and the image data held in the image memory 58, stores it in the image memory 58 again, and outputs it from the terminal 59. Eventually DC
The T-coefficient buffer restores the OCT coefficients from the first term to the 64th term into image data and stores it in the image memory 58. In this way, by sequentially adding code data and finally transferring all the data, the hierarchical restoration for one screen is completed.

[Problem to be solved by the invention]

In the conventional technology, when restoring DCT coefficients to an image,
The DCT coefficients of pixels in all blocks were subjected to inverse DCT/transformation. However, when one block is 8x8 pixels, the inverse DCT transform is an 8x8 matrix operation, and the conversion of one pixel requires eight multiplications and seven additions, that is, the conversion of 64 pixels of l block. requires 512 multiplications and 448 additions. For this reason, when the pixels of all blocks in one screen are subjected to inverse DCT transformation, there is a problem in that it is difficult to speed up image restoration. In particular, when restoring image data (difference data) hierarchically for each DCT coefficient divided into a predetermined number, FIGS.
), many high-order DCT coefficients are zero, and there are many cases where all DCT coefficients in a block to be newly restored are all zero. If all DCT coefficients in the block are zero,
Since the restored image data is also all zero, there is a problem in that a lot of unnecessary processing is performed.

SUMMARY OF THE INVENTION An object of the present invention is to provide an image data restoration device that can increase speed while suppressing the increase in circuits.

[Means to solve the problem]

FIG. 1 is a block diagram of the principle of the present invention.

The present invention divides an original image into blocks consisting of a plurality of N×N images, and quantizes the transform coefficients obtained by performing two-dimensional discrete cosine transform on the tone values of the plurality of N×N images for each block. This is an apparatus for restoring an image from encoded data obtained by encoding quantized coefficients.

The decoding means 1 decodes coded data.

The dequantizing means 2 dequantizes the quantized coefficients decoded by the decoding means 1 to obtain DCT coefficients.

The inverse DCT transform means 3 performs inverse DCT transform on the DCT coefficients dequantized by the inverse quantization means 2.

When there is no significant coefficient among the quantized coefficients in the block decoded by the decoding means 1, the significant coefficient detection means 4 performs inverse quantization by the inverse quantization means 2 and inverse DCT transformation means 3.
Control is performed to output as inverse DCT transformed restored image data.

Image memory 7 stores image data.

The addition means 5 adds the output of the image memory 7 and the output of the inverse DCT transformation means 3. Note that this addition is not performed in the first stage, but is performed after the second stage.

The selection means 6 adds the output of the addition means 5 and the output of the inverse DCT transformation means 3, selects the output of the inverse DCT transformation means 3 in the first stage, and selects the output of the addition means 5 in the second stage and thereafter. output is selected and added to the image memory 7.

[For production]

The input code data is decoded by the decoding means 1 and becomes a quantized coefficient. (2) When there is no significant coefficient among the decoded quantized coefficients within the block, zero data is output as restored data. Further, when there is a significant coefficient, the inverse quantization means 2 and the inverse DCT transformation means 3 are operated to restore the image.

Further, in the first stage, the restored image data is selected and stored in the image memory 7 by the selection means 6, and in the second stage and thereafter, the image data stored in the image memory 7 is inverted.
The image data restored by the CT conversion means 3' is added, and this output is selected by the selection means 6 and stored in the image memory 7.

Further, when the quantization coefficient does not have a significant coefficient within one block, the block address is incremented and addition within the block is not performed.

Thereby, operations such as inverse quantization and addition of inverse DCT transform can be reduced.

〔Example〕

FIG. 2(A) is a configuration diagram of the first embodiment of the present invention.

Code data input from the terminal 11 is input to the variable length decoding section 12. The variable length decoding unit 12 decodes the input code data into fixed length data of an index and a run using a decoding table 13 that is an inverse table to a Huffman table, and outputs the decoded data to the dequantization unit 14. The inverse quantization section 14 determines the number of significant coefficients in the block, and outputs an invalid block signal ZERO to the two-dimensional inverse DCT transformation section 16 if there are no significant coefficients. In addition, if a significant coefficient exists, the input quantization coefficient is multiplied by each value of the quantization matrix 15 to restore the DCT coefficient, and the two-dimensional inverse DC
It is output to the T conversion section 16. When the invalid block signal ZERO is input, the two-dimensional inverse DCT transform unit 16
Invalid block signal ZER without performing dimensional inverse DCT transformation
O is output to terminal 18.

In addition, when the invalid block signal ZERO is not input, the input DCT coefficients are orthogonally transformed by inverse DCT transform, the coefficients of the spatial frequency distribution are converted to image signals, and the terminal 1
Output to 7.

By repeating the above process, 1! The image of plane j is restored.

FIG. 2(B) is an operation flowchart of the present invention. Execution begins when decrypted data is input.

First, variable length decoding processing S1 is executed to obtain decoded data which are quantized coefficients. Then, it is determined whether a significant coefficient exists or not in step S2. If a significant coefficient exists (YES),
Subsequently, the quantized coefficients are dequantized by dequantization processing S3 to obtain DCT coefficients.

The DCT coefficients are then subjected to two-dimensional inverse DCT transformation S4, and a restored signal is outputted S5.

On the other hand, if there is no significant coefficient in discrimination S2 (N
For o), ZERO output is performed as an invalid block signal (
S6), that is, all zero data is output. Processing S5
, After processing S6, it is determined whether all blocks have been completed or not in step S7L, and if not (No), the variable length decoding processing described above is executed again from step S1. Furthermore, when all blocks have been completed (YES), the process ends.

As is clear from the above processing, all Zero data is output when there are no significant coefficients, so inverse quantization processing and two-dimensional DCT transformation processing S3. There is no need to execute S4, and the processing speed can be increased.

FIG. 3 is a block diagram of the invalid block signal generation circuit in the inverse quantization unit 14 according to the embodiment of the present invention, and FIG. 4 is another diagram of the decoded data string.

Hereinafter, calculation of the number of valid coefficients in a block will be explained according to the block diagram of the invalid block signal generation circuit of the present invention shown in FIG.

The index and run decoded data decoded from the code data by the variable length decoding unit 12 are input to the reso multiplexer 21 through the terminal 10 . The demultiplexer 21 alternately selects index and run decoded data from the input decoded data string in accordance with the selection signal (C3L) from the timing control unit 25, and converts the index (IDX) into the DCT coefficient coefficient part 28 run ( RUN) is output to the block end detection section 22.

In the case of a block containing only DC components (FIG. 4(a)), the decoded data (Dl) of the first DC component is sent to the demultiplexer 2.
1 and input to the DCT coefficient part 28. Then, in accordance with the number count signal (ICN) from the timing control unit 25, the effective coefficient number calculation unit 23 counts the number of effective coefficients “1” in the target block. Further, the address (ADR=O) where the decoded data (Dl) exists is held in the effective coefficient selection unit 26. Next, the demultiplexer 21 selects a run (Reob) and outputs it to the block end detection section 22. The block end detection unit 22
Since the input run value (RUN) is (Reob), the coefficients of the remaining pixels of the block are all invalid coefficients,
A signal (
BEN) is output. Then, the timing control unit 25 instructs the effective coefficient number determining unit 24 to determine the number of effective coefficients in the block (BCN). Since the number of effective coefficients in the block is one, the effective coefficient number determination unit 24
It is determined that the target block has a significant coefficient, and the invalid block signal ZERO is output from the terminal 20 as "0".

On the other hand, the DCT coefficient coefficient unit 28 multiplies the held decoded data (Dl) by the quantization threshold held at the DC component address (ADR=O) of the quantized dark value holding unit 27, and The result is output from terminal 29. Then, the two-dimensional inverse DC'T transform unit 16 converts it into an image signal, and "restores" the image signals of all the pixels of the block.

In the case of a program in which DC component and AC component coefficients exist (FIG. 4(b)), first, the decoded data (D2) of the first DC component is selected in the demultiplexer 21, and the D
CT coefficient coefficient part 28 is input.

Then, according to the number count signal (ICN) from the timing control unit 25, the effective coefficient number calculation unit 23 counts the number of effective coefficients “1” in the target block. Further, the effective coefficient selection unit 26 holds the address (ADR=O) where the decoded data (Dl) exists. Next, the demultiplexer 2] selects a run (RO) and outputs it to the block end detection section 22. The input run value is R
O, so the block end detection unit 22 determines that the coefficients within the block have not ended. Effective coefficient selection section 26
Then, the address (ADR) of the next effective coefficient is calculated from the input RUN value and held. On the other hand, since the demultiplexer 21 selects the index (11) this time, the input index (11) is held in the DCT coefficient coefficient unit 28. In addition, the timing control section 25
According to the number count signal ('ICN) from , the number of effective coefficients [+IJ] in the target program is added to the effective coefficient number calculation unit 23, and "2" is counted.

Next, the demultiplexer 21 selects the run (RO),
It is output to the block end detection section 22. Entered run (
RUN) is RO, so the block end detection unit 2
2 determines that the coefficients in the block are not finished. Therefore, the effective coefficient selection unit 26 selects the input run (RUN).
Calculate the address (ADR) of the next effective coefficient from the value of
Hold.

On the other hand, the demultiplexer 21 now outputs the index (
Select I2). The index (I2) input by this selection is stored in the DCT coefficient coefficient unit 28. Further, according to the number count signal (ICN) from the timing control unit 25, the number of effective coefficients in the target block is added to the effective coefficient number calculation unit 23 by “+1”, and the number is “3”.
is counted. Next, the demultiplexer 21 selects a run (Reob) and outputs it to the block end detection section 22. Since the input signal is run (Reob),
The block end detection unit 22 determines that the coefficients of the remaining pixels in the block are all invalid coefficients and that the coefficients in the block have ended, and sends a signal (BEN) to the timing control unit 25 instructing the end of the valid coefficients in the block. Output. Then, the timing control section 25 instructs the effective coefficient number determining section 24 to determine the number of effective coefficients in the block (BCN). Since the number of valid coefficients in the block is three, the valid coefficient number determination unit 24 determines that the target block has significant coefficients, and outputs the invalid block signal ZERO from the terminal 20 as "0". Also, the decoded data (DIII, 12) held in the DCT coefficient coefficient unit 28 is multiplied by the quantized dark value held in the corresponding address (A, DR) of the quantization threshold holding unit 27, and the The result is output from terminal 29. Then, the two-dimensional inverse DCT transform unit 16 converts it into an image signal, and restores the image signals of all pixels of the block.

On the other hand, in the second and subsequent stages of hierarchy restoration, the demultiplexer 21 first selects a run.

If there is no significant coefficient (FIG. 4(C)), the selected run (Reob) is output to the block end detection section 22. Since the input RUN value is Reob, the block end detection unit 22 determines that the coefficients of the remaining pixels in the block are all invalid coefficients and that the coefficients in the block have ended, and the timing control unit 25 A signal (BEN) instructing the end of the valid coefficients in the block is output. Then, the timing control unit 25 instructs the effective coefficient number determining unit 24 to determine the number of effective coefficients in the block (B
CN). Since the number of effective coefficients in the block is 0, the effective coefficient number determination unit 24 determines that there is no significant coefficient in the target block, and outputs an invalid block signal ZERO.
is output from the terminal 20 as "1", and all processing of the block is ended.

FIG. 5 is a block diagram of a second embodiment of the present invention including image updating. First stage code data DI (X01...Xn1:n) divided in advance from the terminal 61
l<64) is decoded by the variable length decoding unit 62 using the decoding table 63 and input to the dequantization unit 64. In the first stage,
Since the DC component is included, the inverse quantization unit 64 sets the invalid block signal to "0" and performs normal inverse quantization processing using the quantization matrix 65. Here, the dequantized DCT
The coefficients are output to the two-dimensional DCT transform unit 66, and are converted into two-dimensional inverse DCT
The CT conversion unit 66 restores it to image data.

The selector 69 is instructed by the signal FIR5T indicating the first stage to select the restored divided image data. At this time, according to the write signal WRITE output from the image memory control section 67, the image memory 68 stores the output of the two-dimensional inverse DCT transformation section 66 selected by the selector 69. Note that this storage address is the address generator 7
1 becomes the address MADR output. By repeating the above processing for all blocks, the first stage image restoration for one screen is completed.

Next, the second stage code data D2 (Xnl+1 ・
...Xn2: nl<n2<64) is restored to image data in the same process as above. However, if there is no significant coefficient as shown in FIG. 4(C), the invalid block signal ZERO is output as "1" as described above. At this time, the image memory control unit 67 does not access the image memory, but only increments the value of the block address generation counter 74 by 1 via the OR circuit 72, as shown in FIG. 5(b). Then, the processing of the block is ended. If a significant coefficient exists, the normally restored image data is added to the previous stage image data in the adding section 70.

The selector 69 selects this addition result, writes it to the image memory 68 according to the write signal WRITE output from the image memory control section 67, and also sets the 6-bit counter 73 that generates the intra-block address by 1 according to the address update request signal REQ. To increase. Then, the write address MADR to the image memory is updated. When all pixels in the block have been updated, the carry signal CAR
RY is output from the 6-bit counter 73, and the OR circuit 7
2 to set the value of the block address generation counter 74 to 1.
, and then ends the processing of the block. By repeating the above processing for all blocks, the second stage image restoration for one screen is completed.

The same processing as the second stage processing is performed on the first stage code data D i (Xni+1 . . . -X64:ni<
By performing steps up to 64), the hierarchy restoration for one screen is completed.

[Effects of the Invention] As explained above, according to the present invention, when there are no effective coefficients in a block, inverse quantization, inverse DCT transformation,
Also, by not updating the image memory, it is possible to realize faster image restoration during hierarchical restoration.

[Brief explanation of the drawing]

Figure 1 is a block diagram of the principle of the present invention, Figure 2 (A) is a configuration diagram of the first embodiment of the present invention, Figure 2 (B) is an operation flowchart of the present invention, and Figure 3 is a block diagram of the present invention. FIG. 4 is a block diagram of the invalid block signal generation circuit of the embodiment, FIG. 4 is another diagram of the decoded data string, FIG. 5 is a block diagram of the second embodiment of the present invention, and FIG. 6 is the ADCT encoding circuit. Block diagram: Figure 7 is a block diagram of the two-dimensional DCT transformer, Figure 8 is A
A block diagram of the restoration circuit of the DCT method, Figure 9 is another diagram of the circuit block of the two-dimensional inverse DCT transformer, Figure 10 is a diagram representing the original image signal, Figure 11 is a diagram representing the DCT coefficients, Figure 12 is a diagram representing the dark value for the DCT coefficient, the 13th
The figure shows quantization coefficients, Figure 14 is an explanatory diagram of the scanning order of quantization coefficients, Figure 15 is a block diagram of a conventional ADCT layer restoration unit, and Figure 16 (A) is an example of division of DCT coefficients ( 1st stage)
, Figure 16 (B) is an example of division of DCT coefficients (second stage)
It is. l...Decoding means, 2...Inverse quantization means, 3...Inverse DCT transformation means, 4...Significant coefficient detection means, 5...Addition means, 6...Selection means, 7. ...Image memory.

Claims (1)

[Claims] 1) Transformation obtained by dividing the original image into blocks each consisting of a plurality of N×N pixels, and performing two-dimensional discrete cosine transformation on the gradation values of the plurality of N×N pixels for each block. A device that quantizes coefficients and restores an image from encoded data obtained by encoding the obtained quantized coefficients, comprising a decoding means (1) for decoding the inputted encoded data, and a quantum decoded by the decoding means (1). an inverse quantization means (2) that dequantizes a quantization coefficient to obtain a DCT coefficient; an inverse DCT transformation means (3) that performs an inverse DCT transformation on the DCT coefficient dequantized by the inverse quantization means (2); When there is no significant coefficient among the quantized coefficients in the block decoded by the decoding means (1), the inverse quantization means (2)
and the inverse DC of the inverse DCT transform means (3).
An image data restoring device comprising significant coefficient detection means (4) for controlling output as restored image data without performing T-transformation. 2) Divide the original image into blocks each consisting of a plurality of N×N pixels, and quantize the transform coefficients obtained by performing two-dimensional discrete cosine transform on the tone values of the plurality of N×N pixels for each block. A device for restoring an image from encoded data obtained by encoding quantized coefficients, comprising: a decoding means (1) for decoding the input coded data; and a dequantizing means for dequantizing the quantized coefficients decoded by the decoding means (1). an inverse DCT transform means (3) that performs inverse DCT transform on the DCT coefficients dequantized by the inverse quantizer (2); and an image that stores image data. a memory (7); an addition means (5) for adding the output of the image memory (7) and the output of the inverse DCT transformation means (3); and the output of the addition means (5) and the inverse DCT transformation means. (3)
and the output of the inverse DCT in the first stage
The output of the converting means (3) is selected, and after the second stage, the output of the adding means (5) is selected and the image memory (7)
selection means (6) for storing the quantization coefficients in the block; and when there is no significant coefficient among the quantized coefficients in the block decoded by the decoding means (1), the inverse quantization means (2)
and the inverse DC of the inverse DCT transform means (3).
The addition means (5
) and significant coefficient detection means (4) for controlling output to the selection means (6). 3) The image memory (7) generates a block address and an intra-block address, and an address that increments the block address when there is no significant coefficient among the quantized coefficients in the block decoded by the decoding means (1). 3. The image data restoration apparatus according to claim 2, further comprising a generating means. 4) Divide the original image into blocks each consisting of a plurality of N×N pixels, and quantize the transform coefficients obtained by performing two-dimensional discrete cosine transform on the tone values of the plurality of N×N pixels for each block. A device for restoring an image from coded data encoded with quantized coefficients, comprising: a decoding means for decoding the input coded data; and a block for detecting the end of the coefficient of the block from the signal decoded by the decoding means. end detection means; effective coefficient number calculation means for counting the number of effective coefficients in the block until the end of the coefficients of the block is detected by the block end detection means; effective coefficient number determining means for determining the number of effective coefficients in a block; and when the effective coefficient number determining means determines that the number of effective coefficients in the block is 0, an invalid block signal is sent. An image data restoration device characterized by outputting. 5) Divide the original image into blocks each consisting of a plurality of N×N pixels, and quantize the transform coefficients obtained by performing two-dimensional discrete cosine transform on the tone values of the plurality of N×N pixels for each block. A device for restoring an image from encoded data obtained by encoding quantized coefficients, which decodes the input coded data into quantized coefficients, detects the presence or absence of a significant coefficient in a block from the decoded signal, and extracts the result. When the result is null, it outputs an invalid block signal and zero data, and when the result is true, it performs inverse quantization to convert the quantized coefficients into DCT coefficients, and then converts the obtained DCT coefficients. An image data restoration method characterized by performing inverse DCT transformation to restore an image signal.