JPH04220081A - Method and device for picture data decoding - Google Patents

Abstract

PURPOSE:To quicken the processing of data without increasing the circuit scale by quantizing a transformation coefficient obtained through 2-dimension discrete cosine transformation (ADCT) to a gradation of plural picture elements in each of plural blocks each comprising plural picture elements (NXN) divided from an original picture and decoding a picture from a coding data resulting from coding the obtained quantization coefficient. CONSTITUTION:In the decoding process comprising a variable length decode means 11 for a coding data, an inverse quantization means 12 and an inverse DCT transformation means 13, when the distribution of a significant coefficient obtained from a DCT coefficient of one block subject to inverse quantization is only a DC component and the DC component is zero, the inverse DCT transformation processing is omitted and all zeros are outputted for picture signals of one block.

Description

[Detailed description of the invention]

0001

[Field of Industrial Application] The present invention relates to an image data restoration method and apparatus for restoring an image from compressed and encoded image data, and in particular, it divides a multivalued image into blocks each consisting of a plurality of pixels, and The present invention relates to an image data encoding/restoring method and apparatus for restoring an image from orthogonal transform encoded data of a multivalued image encoded after orthogonally transforming pixels.

[0002] In order to store or transmit high-speed, high-quality image data, especially halftone and color image data, which has an order of magnitude larger amount of information than numerical data. requires highly efficient encoding of tone values for each pixel. Conventionally, as a highly efficient compression method for image data, there is, for example, an adaptive discrete cosine transform encoding method. Adaptive Discrete Cosine Transform Coding
eCosineTransform Hereinafter, abbreviated as “
(referred to as "ADCT") will be explained next.

[0003] ADCT divides an image into blocks each consisting of 8×8 pixels, and converts the image signal of each block into coefficients of a spatial frequency distribution using two-dimensional discrete cosine transform (hereinafter referred to as "DCT"). Quantization is performed using a threshold value adapted to the quantization coefficient, and the obtained quantization coefficient is encoded using a statistically obtained Huffman table. The encoding operation will be explained in detail according to the basic configuration diagram of ADCT shown in FIG.

First, the image is divided into blocks each consisting of 8×8 pixels as shown in FIG.
Enter 4. The two-dimensional DCT transformation section 24 orthogonally transforms the input image signal using DCT to convert it into DCT coefficients having the spatial frequency distribution shown in FIG.
Output to. Specifically, as shown in FIG.
The input image signal is subjected to one-dimensional DCT transformation in a one-dimensional DCT transformation section 30, and the transposition section 31 transposes the rows and columns of coefficients in the block (transposition), and outputs it to a one-dimensional DCT transformation section 32. In the one-dimensional DCT conversion unit 32, one-dimensional DCT
One-dimensional DCT transformation is performed in the same way as in the transformation unit 30, and the transposition unit 33
Output to. The transposing unit 33 performs the same transposing process as the transposing unit 31 and outputs it to the terminal 34.

[0005] By performing such processing on all blocks of image data, it is converted into DCT coefficients. Figure 5 again
Referring to , the linear quantization unit 25 linearly quantizes the input DCT coefficients using a quantization matrix 29 configured with the threshold values shown in FIG. (hereinafter simply referred to as "quantization coefficient") is obtained. As shown in Figure 12, quantization D
CT coefficients are quantized D, in which DCT coefficients with values smaller than the threshold value are 0, and only DC components and a few AC components have values.
CT coefficients are generated.

The two-dimensionally arranged quantized DCT coefficients are converted into one-dimensional data according to a scanning order called zigzag scan shown in FIG. The variable length encoding unit 26 encodes the DC at the beginning of each block.
The difference between the component and the DC component of the previous block is variable-length encoded. For AC component, effective coefficient (coefficient whose value is not 0)
The value of (hereinafter referred to as "index") and the length of the run (hereinafter referred to as "run") of invalid coefficients (coefficients with a value of 0) up to that point are variable-length coded for each block. D.C.
, AC components are encoded using a code table 27 consisting of a Huffman table created based on statistics for each image, and the obtained coded data is sequentially output from a terminal 28. Ru.

On the other hand, encoded data is restored to an image by the following method. FIG. 7 shows a block diagram of an ADCT restoration circuit, and FIG. 8 shows a block diagram of a two-dimensional inverse DCT transform section. Figure 7
, code data input from the terminal 40 is input to the variable length decoding section 41. In the variable length decoding unit 41, the input code data is converted into fixed length data of index and run using a decoding table 42 which is an inverse table to the Huffman table of the code table 27 in FIG. It is decoded and output to the inverse quantization section 43. The dequantization unit 43 dequantizes the input quantization coefficients by multiplying them by each of the quantization matrices 29 to restore the DCT coefficients, and performs two-dimensional inverse DCT
It is output to the converter 44.

The two-dimensional inverse DCT transform section 44 orthogonally transforms the input DCT coefficients by inverse DCT transform, and transforms the coefficients of the spatial frequency distribution into an image signal. Specifically, as shown in FIG. 8, the DCT coefficient input from the terminal 50 is 1.
A one-dimensional inverse DCT transform unit 51 performs a one-dimensional inverse DCT transform, and the result is output to a transpose unit 52 . The transposing unit 52 transposes the rows and columns of coefficients within one block and outputs the result to the one-dimensional inverse DCT transform unit 53.

The one-dimensional inverse DCT transform unit 53 again performs one-dimensional inverse DCT transform on the input transposed coefficients, and the transpose unit 54
Output to. The transposing unit 54 again performs 1 in the same way as the transposing unit 52.
The image is restored by interchanging the rows and columns of the coefficients within the block and outputting the obtained signal from the terminal 45.

0010

2. Description of the Related Art FIG. 14 shows an overall block diagram of a conventional image data restoration device. In FIG. 14, the variable length decoding section 41, decoding table 42, inverse quantization section 43, quantization matrix 29, and two-dimensional inverse DCT transformation section 44 are the same as those in FIG. An image memory 60 is provided.

[0011] The code data is sent from the terminal 40 to the variable length decoding unit 4.
1 is input. The scanning order of the DCT coefficients is set as X01, X02,...X64 according to the zigzag operation order in Figure 13.
Then, the variable length decoding unit 41 decodes the code data corresponding to the DCT coefficients of the first term X01 to the 64th term X64 into quantized DCT coefficients. The decoded quantized coefficients are output to the dequantization section 43 and are dequantized into DCT coefficients. The inverse quantized DCT coefficients are converted into two-dimensional inverse DCT
The converted image data is converted into image data by the conversion unit 44, and the obtained image data is held in the image memory 60 and outputted from the terminal 61, thereby restoring an image for one screen.

0012

[Problems to be Solved by the Invention] In such a conventional image data restoration device, when restoring DCT coefficients to an image, the DCT coefficients of pixels of all blocks are subjected to inverse DC conversion.
It is T-converted. However, inverse DCT transformation is an 8x8 matrix operation when one block is 8x8 pixels, and the conversion of one pixel requires eight multiplications and eight additions, that is, the conversion of 64 pixels of one block. requires 512 multiplications and 512
It is necessary to add times. 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.

[0013] When a high-speed search such as a database search is required, a hierarchical restoration method is used in which a rough image of poor quality is restored at an early stage and the image quality is gradually improved. In such hierarchical restoration, even if the amount of code is reduced by configuring the first layer used for image search only with DC components as shown in FIG. Since the inverse DCT coefficients are distributed in all of the 1st to 64th terms of the DCT coefficients, there is a problem that images cannot be restored at high speed.

The present invention has been made in view of these conventional problems, and it is an object of the present invention to provide an image data restoration method and apparatus that can increase the speed without increasing the circuit scale.

[0015]

[Means for Solving the Problems] FIG. 1 is an explanatory diagram of the principle of image data restoration according to the present invention. First, in the present invention, for each block obtained by dividing an original image into a plurality of blocks each consisting of a plurality of pixels (N x N), the tone values of the plurality of pixels in the block are divided into two-dimensional discrete The present invention is directed to an image data restoration device that quantizes transform coefficients obtained by cosine transformation and restores an image from encoded data obtained by encoding the obtained quantized coefficients.

In the present invention, such an image data restoring device includes variable length decoding means 11 for decoding quantized coefficients from input coded data; Inverse quantization means 12 that inversely quantizes DCT coefficients; Inverse DCT transformation means 13 that performs inverse DCT transformation on the DCT coefficients obtained by the inverse quantization means 12 to restore an image signal; Determine the number of significant coefficients from the quantized DCT coefficients of one block,
DC component block determining means 14 determines that the significant coefficient is one and is a DC component; DC component block determining means 14 determines that the significant coefficient in the block is only a DC component DC component zero detection means 15 for detecting whether or not the component is zero; When the DC component zero detection means 15 determines that the DC component in a block is zero, all pixel signals of one block are set to zero. It is characterized by outputting an invalid block signal. The method for restoring image data according to the present invention includes a first step of decoding quantized coefficients from input coded data;
a second process of inversely quantizing the quantized coefficients decoded in the first process into DCT coefficients; a third process of inversely DCT-transforming the DCT coefficients obtained in the second process and restoring them to image signals; A fourth step of detecting the distribution of significant coefficients from one block of DCT coefficients dequantized in the process; From the distribution of significant coefficients detected in the fourth step, it is determined whether the value of the DC component is zero when only the DC component exists and a sixth step of selecting whether or not to perform inverse DCT transformation in the third step according to the detection result of the fifth step. In this sixth process, it is determined that the significant coefficient is only a DC component in the fourth process,
In addition, if a block in which the DC component is zero is detected in the fifth step, the inverse DCT transform process in the third step is omitted and all zeros are output as one block image signal, or an invalid block signal is output. Output.

[0017]

[Operation] According to the image data restoration method and apparatus of the present invention having such a configuration, it is possible to utilize the fact that image data obtained by performing inverse DCT transformation of a block whose DCT coefficients are all zeros are all zeros. By skipping the inverse DCT transformation process for blocks where the only significant coefficient among the restored DCT coefficients is the DC component and the value of the DC component is zero, The number of calculations required for the restoration process can be significantly reduced, and the average image restoration speed can be improved with a simple circuit.

[0018]

Embodiment FIG. 2 is a block diagram showing an embodiment of the present invention. Incidentally, the correspondence with the embodiments is indicated by the numbers obtained by adding 100 to the numbers in the principle explanatory diagram of FIG. In FIG. 2, 111 is a variable length decoding unit, 112 is an inverse quantization unit, 1
09 is a quantization matrix, 113 is a two-dimensional inverse DCT transform unit, 114 is a DC block determination unit, 115 is a DC zero detector, 116 is an address generation unit, 117 is an image memory control unit, and 118 is an image memory. The image memory control unit 117 also includes a block address generation counter 121, a 6-bit counter that generates 64 addresses in a block, and an OR circuit 123, as shown in FIG.

Next, a processing operation for restoring image data will be explained. Code data input from terminal 110 is input to variable length decoding section 111 . The variable length decoding unit 111 decodes the input code data into fixed length data of index and run, and outputs it to the dequantization unit 112. The inverse quantization unit 112 converts the input quantization coefficients into a quantization matrix 1
The DCT coefficients are multiplied by a quantization threshold of 0.09 to restore the DCT coefficients, and are output to the two-dimensional inverse DCT transform unit 113 and also to the DC block determination unit 114.

The DC block determination unit 114 receives the input 1
Determine the number of significant coefficients of the DCT coefficients of the block. Depending on this determination result, the following processing is performed. (1) When it is determined that the significant coefficient in the block is only a DC component, the DC block determination unit 114 sends a signal DCM indicating that the block has only a DC component together with a DC component signal.
is turned on and output to the DC zero detector 115. The DC zero detector 115 detects the input D when the DCM signal is on.
Determine the value of the C component. As a result, if the value of the DC component is "zero", a DC zero block signal DCZ indicating that all data in the block is zero is turned on and output to the address generation section 116 and the image memory control section 117.

When the DC zero block signal DCZ is on, the image memory control unit 117 does not access the image memory 118, but uses the DC zero block signal DCZ via the OR circuit 86 as shown in FIG. By simply increasing the value of the block address generation counter 121 by 1,
Finish processing this block. On the other hand, DC zero detector 1
If the value is "non-zero" as a result of the determination in step 15, the input DCT coefficients are restored to image data in the two-dimensional inverse DCT transform unit 113, and are stored in the image memory according to the write signal WRITE output from the image memory control unit 117. 118
At the same time, the address update request signal REQ causes the 6-bit counter 122 in FIG. When all pixels in the block have been updated, a carry signal CARRY is output from the 6-bit counter 122, which increments the value of the block address generation counter 121 by 1 via the OR circuit 123, and the processing of this block ends. do. (2) If it is determined that the significant coefficients in the block are a DC component and an AC component, the DC block determination unit 114 turns off the DC zero block signal DCM. As a result, the input DCT coefficients are restored to image data in the two-dimensional inverse DCT transform section 113, and the image memory 118 is restored according to the write signal WRITE output from the image memory control section 117.
At the same time, the address update request signal REQ causes the 6-bit counter 122 in FIG.

When all pixels in the block have been updated, the carry signal CARRY is sent to the 6-bit counter 122.
The value of the block address generation counter 121 is incremented by 1 via the OR circuit 123, and the processing of this block is completed. By repeating the above processing for all blocks, one screen of images is restored.

FIG. 4 is a flow chart showing the processing operation of the embodiment shown in FIG.
The method for restoring image data according to the present invention is clear from the following. In the embodiment of the present invention, the DC zero detector 115
When the component is determined to be "non-zero", the image data is restored by the two-dimensional inverse DCT transformer 113, but the two-dimensional inverse DCT transformer 113 does not perform the inverse transform and converts the image data to "non-zero".
The image data within the block may be uniquely restored based on the determined value of the DC component.

That is, if the DCT coefficient has only a DC component and D
When the value of the C component is "non-zero", a value obtained by shifting the value of the DC component by 3 bits and assigning it to all pixels in the block matches the result of the secondary inverse DCT transformation. Therefore,
When the DC component is determined to be "non-zero" by the DC zero detector 115, a process is performed in which a value obtained by shifting the value of the DC component by 3 bits is commonly assigned to 64 addresses in the block as a pixel signal, and a restoration process is performed. Speed up.

Furthermore, in the above embodiment, the DC block determination unit 114 performs the determination as to whether or not the significant coefficient in the block to be restored is only the DC component and the zero detection of the DC component after inverse quantization.
Although the case is shown in which the detection is performed by the DC zero detector 115, the variable length decoding section 111 or the inverse quantization section 112 may also perform the processing.

[0026]

As explained above, according to the present invention, when the significant coefficient in a block is only a DC component and the DC component is zero, pixels in the block are transformed without performing inverse DCT transformation. By setting all the signals to zero, it is possible to realize faster image restoration.

[Brief explanation of the drawing]

[Fig. 1] Diagram explaining the principle of image data restoration of the present invention

[Figure 2]
Embodiment configuration diagram of an image data restoration device of the present invention

[Fig. 3] Embodiment configuration diagram of the image memory control unit in Fig. 2.

FIG. 4 is a flowchart showing data restoration processing according to the embodiment of FIG. 2;

[Figure 5] Block diagram of a conventional ADCT encoding circuit

[Figure 6] Block diagram of the two-dimensional DCT transform unit in Figure 5

[Figure 7
]Block diagram of conventional ADCT method restoration circuit

[Figure 8]
Block diagram of the two-dimensional inverse DCT transform unit in Figure 7

[Fig. 9] Explanatory diagram of the original image signal for one block

[Fig. 10] Explanatory diagram of DCT coefficients obtained from Fig. 9

[Fig. 11] Explanatory diagram of quantization threshold used for quantization of DCT coefficients

[Figure 12] Explanatory diagram of quantized coefficients obtained by quantizing DCT coefficients using a quantization threshold

[Fig. 13] Explanatory diagram showing the scanning order of quantization coefficients

[Figure 1
4] Block diagram of conventional image data restoration device

[Figure 15]
DCT coefficient explanatory diagram of only the DC component in the first stage in hierarchical restoration

[Explanation of symbols]

11: Variable length decoding means 12: Inverse quantization means 13: Inverse DCT transformation means 14: DC component block discrimination means 15: DC component zero detection means 109: Quantization matrix 111: Variable length decoding section 112: Inverse quantization section 113 : Two-dimensional inverse DCT conversion section 114: DC block determination section 115: DC zero detector 116: Address generator 117: Image memory control section 118: Image memory 121: Block address generation counter 122: 6-bit counter 123: OR circuit

Claims (4)

[Claims] Claim 1: Each original image has a plurality of pixels (N×N)
For each block obtained by dividing the blocks into a plurality of blocks, the gradation values of the plurality of pixels in the block are subjected to two-dimensional discrete cosine transform, and the transform coefficients obtained are quantized, and the obtained quantization In an image data restoration method for restoring an image from coded data obtained by encoding coefficients, a first step is to decode quantized coefficients from input coded data, and a DCT is performed on the quantized coefficients decoded in the first step. A second process of inverse quantization into coefficients, and an inverse DCT coefficient of the DCT coefficients obtained in the second process.
a third step of performing CT conversion and restoring to an image signal; and the second step of
A fourth step of detecting the distribution of significant coefficients from one block of DCT coefficients dequantized in the process, and a step of detecting the distribution of significant coefficients from the distribution of significant coefficients detected in the fourth step, and detecting that the value of the DC component is zero when there is only a DC component. An image data restoring method comprising: a fifth step of detecting whether or not the above-mentioned is the case; and a sixth step of selecting whether or not to perform the inverse DCT transformation in the third step according to the detection result of the fifth step. 2. The image data restoration method according to claim 1, wherein in the sixth step, the significant coefficient is determined to be only a DC component in the fourth step, and the DC component is determined to be zero in the fifth step. An image data restoration method characterized in that when a certain block is detected, the third step of inverse DCT transformation processing is omitted and all zeros are output as an image signal of one block. 3. The image data restoration method according to claim 1, wherein in the sixth step, the significant coefficient is determined to be only a DC component in the fourth step, and the DC component is determined to be zero in the fifth step. An image data restoring method characterized in that when a certain block is detected, the inverse DCT transform in the third step is not performed and an invalid block signal output from the fifth step is selected. Claim 4: Each original image has a plurality of pixels (N×N)
For each block obtained by dividing the blocks into a plurality of blocks, the gradation values of the plurality of pixels in the block are subjected to two-dimensional discrete cosine transform, and the transform coefficients obtained are quantized, and the obtained quantization An image data restoration device that restores an image from code data obtained by encoding coefficients, and variable length decoding means (11) that decodes quantized coefficients from input code data.
and an inverse quantization means (12) for inversely quantizing the quantized coefficients decoded by the variable length decoding means (11) into DCT coefficients; Inverse DCT transformation means (
13) and a DC component block that determines the number of significant coefficients from one block of DCT coefficients dequantized by the dequantization means (12), and determines that there is one significant coefficient and that it is a DC component. For a block in which the determining means (14) and the DC component block determining means (14) determine that the significant coefficient within the block is only a DC component,
DC component zero detection means (1) for detecting whether the DC component is zero or not
5), and when the DC component zero detection means (15) determines that the DC component in the block is zero, outputting an invalid block signal that makes all the pixel signals of one block zero. Characteristic image data restoration device.