JPH05292326A - Picture data coder - Google Patents

Abstract

PURPOSE:To allow the picture data coder to generate code data from which all pictures with a different gradation change characteristic are decoded with high picture quality. CONSTITUTION:A 2-dimension DCT coefficient of each block outputted from a 2-dimension DC transformation section 151 is quantized at a linear quantization section 152 by using a quantization condition inputted from a quantization condition decision section 180. A picture quality discrimination 160 uses a quantization coefficient or the like of each block obtained through the quantization to calculate an evaluation function to discriminate the deterioration degree of a decoded picture based on the evaluation function obtained by the arithmetic operation and outputs a quantization condition to decrease the quantization error from the quantization condition decision section 180 to the linear quantization section till the degree of the deterioration in the picture quality reaches a prescribed degree or below. When the picture deterioration degree reaches a permissible degree or below, a coding command is outputted to a variable length coding section 154.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image data coding apparatus for coding multi-valued image data with high efficiency, and more specifically, it divides multi-valued image data into a plurality of blocks, and each divided block is divided into blocks. The present invention relates to an image data coding device that generates code data for image restoration by using the orthogonal transform coefficient of each block obtained by performing the orthogonal transform on the.

[0002]

2. Description of the Related Art In the fields of video conferencing / videophones, color still images, color moving images, etc., image data having an order of magnitude greater than numerical data, especially halftone images and color image data, is stored. Furthermore, since it is necessary to transmit the image data at high speed and with high quality, it is essential to perform a process of efficiently encoding the gradation value of each pixel.

An adaptive discrete cosine transform coding method is known as a highly efficient compression method for such image data. This Adaptive Discrete Cosine Transform method, hereinafter,
The abbreviated method is called ADCT method).
It is divided into a plurality of blocks each consisting of × 8 pixels, and the image signal of each block is converted into a coefficient of spatial frequency distribution (two-dimensional DCT coefficient) by two-dimensional discrete cosine transform (hereinafter referred to as two-dimensional DCT). A two-dimensional DCT coefficient in each block obtained by the transformation is quantized into a threshold value adapted to the visual sense, and the quantized coefficient obtained by the quantization is encoded by a Huffman table obtained statistically. Is.

Next, the encoding operation of this ADCT method will be described.
This will be described in more detail with reference to the basic block diagram of the conventional image data encoding device shown in FIG. First, the multi-valued image data (original image signal) is divided into a plurality of blocks of 8 × 8 pixels as shown in FIG. 14, and the image signal 150 of each block is sequentially input to the two-dimensional DCT conversion section 24. The values shown in FIG. 14 are gradation level values of each pixel in one block.

The activity calculation unit 156 calculates the total sum of the differences between the gradation level value of each pixel in the block and the average gradation level value in the block, and outputs the total value to the quantization condition selection unit 153 as an activity. To do. The quantizing condition selecting unit 153 selects the most suitable quantizing condition from a plurality of quantizing conditions according to the magnitude of the input activity value, and sets the quantizing condition to the linear quantizing unit 15
Output to 2.

On the other hand, in the two-dimensional DCT conversion section 151, the input one-block image signal 150 is orthogonally transformed by the two-dimensional DCT, and for example, one-block image signal as shown in FIG. 14 is shown in FIG. 8 × 8 with various spatial frequency distribution
The two-dimensional DCT coefficients are converted and output to the linear quantization unit 152. The value of each cell shown in FIG. 15 is a two-dimensional DC.
It is the value of the T coefficient.

The linear quantizing unit 152 determines the 8 × 8 two-dimensional DCT coefficients for each block inputted as described above by a visual experiment inputted from the quantizing condition selecting unit 153, for example, Linear quantization is performed by the quantization matrix 157 configured with threshold values as shown in FIG. The result obtained by this linear quantization is shown in FIG. As shown in the figure, the two-dimensional DCT coefficient having a value smaller than the threshold is 0.
Therefore, the DC component (= 5) and the slight AC component (=-
8 × 8 quantized coefficients whose values are only 2, -3,1,1, -1) are generated.

Next, the 8 × 8 quantized coefficients arranged two-dimensionally as shown in FIG. 17 are scanned by a so-called zigzag scan for scanning the quantized coefficients in the order of the numbers shown in FIG. It is converted into a one-dimensional numerical sequence and input to the variable length coding unit 154. The variable length coding unit 154 performs variable length coding on the difference between the DC component at the beginning of each block and the DC component at the beginning of the previous block. For the AC component, the run length (hereinafter, referred to as zero run length) of the value of the effective coefficient (coefficient whose value is not 0) (hereinafter, referred to as index) and the invalid coefficient (coefficient whose value is 0) up to that value are called. ) And are combined, and variable length coding is performed for each block. In this variable length coding, each of DC and AC components is a Huffman table that is created based on a statistic for each image.
Coded data 158 obtained by the coding using 55 is sequentially output to the outside.

[0009]

As described above, in the conventional ADCT method, when the two-dimensional DCT coefficient of one block (8 × 8) is quantized, all the two-dimensional DCT coefficients in the block are set to one. Types of quantization conditions (quantization matrix 1
57) is used for quantization.

By the way, human visual characteristics tend to make it relatively difficult to recognize deterioration of image quality for an image whose gradation changes drastically even if there are some approximation errors in the gradation values of pixels. is there.

The quantizing matrix 157 shown in FIG. 16 which is selectively output from the quantizing condition selecting unit 153 utilizes this human visual characteristic. On the other hand, human visual characteristics have a strong discriminating power for an approximation error with respect to an image in which gradation changes gradually change, and tend to easily recognize image quality deterioration.

Therefore, when quantization is performed using the quantization matrix 157 suitable for an image with a sharp gradation change, when an image with a gentle gradation change is restored, the deterioration of the image becomes noticeable. ..

In order to prevent such image deterioration,
Each value of the quantization matrix 157 may be set to a low value to improve the accuracy of quantization. However, in this case, the number of effective coefficients increases in an image in which gradation changes drastically, so that a new problem arises that the amount of code at the time of encoding increases and efficient image compression cannot be performed. ..

An object of the present invention is to enable all images to be coded so as to obtain high image quality regardless of the degree of gradation change.

[0015]

FIG. 1, FIG. 2, FIG. 3, and FIG. 4 are principle block diagrams of the first, second, third, and fourth inventions, respectively. These inventions divide the image data of the original image into a plurality of blocks, perform an orthogonal transform on the gradation values of a plurality of pixels in each of the divided blocks, and then It is premised on an image data coding apparatus that quantizes a plurality of orthogonal transform coefficients of each block obtained by the orthogonal transform and then generates code data for image restoration.

First, the first invention shown in FIG. 1 comprises the following respective means. The quantization condition output means 1 outputs a plurality of quantization conditions used for quantizing a plurality of orthogonal transform coefficients in each block.

The quantizing means 2 quantizes a plurality of orthogonal transform coefficients in each block by using the quantizing condition from the quantizing condition outputting means 1. The encoding means 3 encodes the quantized coefficient generated by the quantization means 2.

The coding determination means 4 is restored using the quantized coefficient in each block quantized by the quantization means 2 and the quantization condition used by the quantization means 2 for the above quantization. Using a plurality of orthogonal transform coefficients in each block and the orthogonal transform coefficient before quantization, a predetermined evaluation function for evaluating the deterioration degree of the image of each block is calculated, and the evaluation function obtained by the calculation is performed. Depending on the value of, it is determined whether the degree of deterioration of the restored image of each block is within the allowable range,
When the degree of deterioration of the restored image of each block exceeds the allowable range, the quantization condition output means 1 is instructed to output the quantization condition such that the quantization error becomes smaller than the previous quantization. On the other hand, if the degree of deterioration of the restored image of each block is within the allowable range, the encoding means 3 is instructed to perform encoding.

Next, the second invention shown in FIG. 2 comprises the following respective means. The quantization condition output means 11 outputs a plurality of quantization conditions used for quantizing a plurality of orthogonal transform coefficients in each block.

The quantizing means 12 is the quantizing condition outputting means 1
Using the quantization conditions from 2, the plurality of orthogonal transform coefficients in each block are quantized for each block. The encoding unit 13 encodes the quantized coefficient generated by the quantization unit 12.

The image quality judging means 14 is restored by using the quantized coefficient in each block quantized by the quantizing means 12 and the quantizing condition used by the quantizing means 12 for the quantizing. Using a plurality of orthogonal transform coefficients in each block and the orthogonal transform coefficient before quantization, a predetermined evaluation function for evaluating the deterioration degree of the image of each block is calculated, and the evaluation function obtained by the calculation is performed. It is determined whether or not the value of satisfies a predetermined condition.

The counting means 15 counts the number of blocks in the image data of the original image, which satisfies the above-mentioned predetermined condition, based on the judgment result of the image quality judging means 14. When the number of blocks counted by the counting unit 15 is smaller than a predetermined number, the encoding determination unit 16 determines a quantization condition that causes a quantization error for the quantization condition output unit 11 to be smaller than the previous quantization. On the other hand, if the value of the evaluation function satisfies the predetermined condition, the encoding means 13 is instructed to perform encoding.

Subsequently, the third invention shown in FIG. 3 comprises the following means. The quantization condition output means 21 outputs a plurality of quantization conditions used for quantizing a plurality of orthogonal transform coefficients in each block.

The quantizing means 22 is the quantizing condition outputting means 2
Using the quantization conditions starting from 1, a plurality of orthogonal transform coefficients in each block are quantized for each block. The encoding unit 23 encodes the quantized coefficient generated by the quantization unit 21.

The evaluation average value calculating means 24 uses the quantized coefficient in each block quantized by the quantizing means 22 and the quantizing condition used by the quantizing means 22 for the quantizing, By using a plurality of orthogonal transform coefficients in each restored block and an orthogonal transform coefficient before quantization, an evaluation function of all blocks is calculated by performing a calculation of a predetermined evaluation function for evaluating the degree of deterioration of the image of each block. The value is obtained, and the average value of the evaluation function values of one block in the entire image is obtained from the evaluation function values of all the blocks.

The encoding judging means 25 uses the average value obtained by the evaluation average value calculating means 24 to quantize the quantizing error with respect to the quantizing condition outputting means 21 so as to be smaller than the previous quantizing. Either the output instruction of the encoding condition or the encoding instruction to the encoding means 23 is performed.

Finally, the fourth invention shown in FIG. 4 comprises the following means. Quantization condition output means 31
Outputs a plurality of quantization conditions used to quantize a plurality of orthogonal transform coefficients in each block.

The quantizing means 32 is the quantizing condition outputting means 3
Using the quantization conditions starting from 1, a plurality of orthogonal transform coefficients in each block are quantized for each block. The encoding unit 33 encodes the quantized coefficient generated by the quantization unit 32.

The evaluation function calculation means 34 is the quantization means 32.
The quantized coefficient in each block quantized by
Using the quantization conditions used by the quantizing means 32 for the above-mentioned quantization, a plurality of orthogonal transform coefficients in each block to be restored and the orthogonal transform coefficient before quantization are used to deteriorate the image of each block. A predetermined evaluation function for evaluating the degree is calculated, and an evaluation function value indicating that the quantization error is the largest among the evaluation function values obtained by the calculation is obtained.

The coding judgment means 35 is based on the evaluation function value obtained by the evaluation function calculation means 34 so that the quantization error for the quantization condition output means 31 is smaller than that of the previous quantization. Either the output instruction of the quantization condition or the encoding instruction to the encoding means (33) is given.

The above-mentioned quantization condition is, for example, a matrix (quantization matrix) made up of a plurality of quantization threshold values set in a one-to-one correspondence with respect to each orthogonal transform coefficient in the orthogonal transform coefficient block. Is.

[0032]

In the first aspect of the invention, the quantization condition input from the quantization condition output unit 1 is used every time the quantization unit 2 inputs a plurality of orthogonal transform coefficients of each divided block of the original image. , The plurality of orthogonal transform coefficients of each block are quantized to generate a plurality of quantized coefficients corresponding to the respective orthogonal transform coefficients.

The coding determination means 4 determines a plurality of orthogonal transform coefficients of each block, the quantization condition, and a plurality of quantized coefficients corresponding to the plurality of orthogonal transform coefficients obtained by the quantization. Perform the calculation of the evaluation function as a variable,
Based on the value of the evaluation function obtained by the calculation, the degree of deterioration of the restored image of each block is judged, and if the degree of deterioration exceeds the allowable range, the quantization condition output means 1 is given more than the previous time. The quantizing means 2 is instructed to output the quantizing condition that reduces the quantizing error.

The quantizing condition output means 1 outputs a quantizing condition satisfying the instructed condition in response to this instruction, and the quantizing means 2 performs the above quantizing again according to the quantizing condition. This quantization process is repeated until the encoding determination unit 4 determines that the degree of deterioration of the restored image of each block is within the allowable range. Then, when the encoding determination means 4 determines that the degree of deterioration of the restored image of each block is within the allowable range, it instructs the encoding means 3 to perform encoding, and the encoding means 3 receives the instruction. The quantized coefficient generated by the quantization means 2 is encoded.

As described above, in the first aspect of the invention, the degree of deterioration of the restored image using the quantization coefficient generated by the quantization means 2 is checked for each divided block of the original image, and the image quality of a predetermined level or more is checked. The image data for image restoration is encoded by using the quantization condition that allows the image to be restored.

Therefore, it is possible to generate code data that can be restored and displayed with high image quality for all images having different display characteristics such as different degrees of gradation change. next,
In the second invention, the quantizing means 12 uses the quantizing condition output from the quantizing condition outputting means 11 in the same manner as in the first invention, so that a plurality of orthogonal transforms of respective divided blocks of the original image are performed. A quantized coefficient is generated for the coefficient.

The image quality determination means 14 corresponds to each of the plurality of orthogonal transform coefficients of each divided block and each of the orthogonal transform coefficients obtained by the quantization, each time the quantization means 12 performs the quantization. From the quantized coefficient and the quantized condition used for the above quantization, a predetermined evaluation function with those variables is calculated, and the restored image is deteriorated for each divided block from the evaluation function value obtained by the calculation. Judge whether the degree is within the allowable range.

The counting means 15 counts the number of blocks for which the image quality judging means 14 judges that the degree of deterioration of the restored image exceeds the allowable range each time the quantizing means 12 generates the quantized coefficient of each divided block. I will do it.

When the quantizing means 12 completes the quantization of all the divided blocks of the original image, the coding judging means 16 determines that the number of blocks whose deterioration degree of the restored image exceeds the permissible range from the counting means 15. To see if
If the number exceeds the allowable range, the quantization condition output means 11
On the other hand, the quantizing means 12 is instructed to output the quantizing condition such that the quantizing error becomes smaller.

In response to this, the quantizing means 12 restarts the quantizing for all the divided blocks of the original image using the newly inputted quantizing condition, and thereafter, the same operation as described above is repeated.

This repetition of the quantization is repeated until the coding judgment means 16 judges that the number of divided blocks in which the degree of deterioration of the restored image exceeds the allowable range is equal to or less than the predetermined allowable number, and the coding means 16 is repeated. Instructs the encoding means 13 to perform encoding when the number of divided blocks in which the degree of deterioration of the restored image exceeds the allowable range and is equal to or smaller than the allowable number.

Upon receiving this instruction, the coding means 13 codes the quantized coefficients of all the divided blocks of the original image generated by the quantizing means 12. As described above, in the second invention, the orthogonal transformation coefficient is quantized for all the divided blocks of the original image, and the deterioration degree of the image restored using the quantized coefficient obtained by the quantization is within the allowable range. When the number of blocks to be exceeded is less than or equal to the allowable number, the quantized coefficients of all the divided blocks of the original image are encoded.

Therefore, similar to the first invention, for all images having different display characteristics such as different gradation changes,
It is possible to generate code data that can be restored and displayed with high image quality.

Next, in the third invention, the quantizing means 22 uses the quantizing condition inputted from the quantizing condition outputting means 21 to quantize the orthogonal transform coefficients for all the divided blocks of the original image. To do.

The evaluation average value calculating means 24 is the quantizing means 2
2 is quantized in each divided block unit,
Orthogonal transform coefficients of each divided block to be quantized, quantized coefficients corresponding to the individual orthogonal transform coefficients of each of the divided blocks generated by the quantizing unit 12, and quantized for the above quantization. A predetermined evaluation function is calculated by using the quantization condition used by the means 12 as a variable, and the evaluation values obtained by the calculation are sequentially accumulated. Then, when the quantization unit 22 completes the quantization of all the divided blocks of the original image, the evaluation average calculation unit 24 calculates the sum of the evaluation function values of all the divided blocks of the original image obtained by the accumulation as the number of divided blocks. The average of the evaluation function values per block is obtained by dividing by, and the deterioration degree of the image quality of the entire image when the image is restored using the quantization coefficient obtained by the above quantization by the average value of the evaluation function If the deterioration degree of the image quality exceeds the allowable range, the quantization condition output means 21 is instructed to output to the quantization means 22 a quantization condition that reduces the quantization error. .. Then, the same operation as described above is performed again using the new quantization condition.

In the process of generating the quantized coefficients of all the divided blocks of the original image, the coding judgment means 23 says that the quantized coefficients of which the degree of deterioration of the entire restored image is within the allowable range are generated for all the divided blocks. The process is repeated until the condition is satisfied, and when the above condition is satisfied, the encoding determination means 2
3 instructs the encoding means 25 to perform encoding.

In response to this instruction, the encoding means 25 encodes the quantized coefficients of all the divided blocks of the original image generated by the quantization means 22. Thus, in the third aspect of the invention, as in the second aspect of the invention, every time the quantizing means 22 completes the quantization of all the divided blocks of the original image,
Whether or not the deterioration degree of the image restored using the generated quantization coefficient is within the allowable range is determined based on the average deterioration degree of the entire image, and when it is within the allowable range, the quantizing means 22 The quantized coefficients of all the divided blocks of the generated original image are encoded.

Therefore, the same effect as the first and second inventions can be obtained. Finally, in the fourth invention, as in the second and third inventions, the quantization of the orthogonal transform coefficients of all the divided blocks of the original image is performed by the quantization means 32 from the quantization condition output means 31. Using the output quantization condition, the division blocks are successively performed in a predetermined order.

During these operations, the evaluation function calculating means 34
Every time the quantizing means 32 completes the quantization of each divided block, the orthogonal transform coefficient in the divided block that is the target of the quantization and the quantization in the divided block obtained by the above quantization. A coefficient and a quantization condition used for the above quantization are calculated as variables and a predetermined evaluation function. Then, the evaluation function calculation means 34 compares the value of the newly obtained evaluation function with the value of the previously obtained evaluation function each time the calculation of the evaluation function is completed, and confirms that the quantization error is larger. The evaluation function value shown is retained.

As a result, when the quantizing means 32 completes the quantization of all the divided blocks of the original image, the evaluation function calculating means 34 evaluates the divided blocks having the largest quantization error in the quantization. Holds the function value.

The coding judging means 35 inputs the evaluation function value of the quantized divided block having the largest quantization error from the evaluation function calculating means 34, and based on the evaluation function value, the divided block is inputted. It is judged whether the degree of deterioration of the restored image is within the allowable range, and if it exceeds the allowable range,
The quantizing condition output means 31 is instructed to output the quantizing condition to the quantizing means 32 so that the quantizing error becomes smaller.

When the new quantizing condition is input, the quantizing means 32 performs the above-mentioned quantizing process again,
The same operation as described above is repeated again. The series of operations associated with the new quantization is repeated until the encoding determination unit 35 determines that the degree of deterioration of the restored image of the quantized divided block having the largest quantization error is within the allowable range. When the encoding determination unit 35 determines that the degree of deterioration of the restored image of the quantized divided block with the largest quantization error is within the allowable range, it instructs the encoding unit 33 to perform encoding.

In response to this instruction, the coding means 33 codes the quantized coefficients of all the divided blocks of the original image generated by the quantizing means 22. As described above, according to the fourth aspect of the invention, the deterioration degree of the divided block having the largest deterioration degree when the image is restored is used as a reference, and the deterioration degree of the original image is determined when the deterioration degree is within a predetermined allowable range. Encode the quantized coefficients of all divided blocks.

Therefore, the same effect as the first, second and third inventions can be obtained.

[0055]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 5 is a system block diagram of an image data encoding apparatus according to an embodiment of the present invention. In the same figure, the same name and code are given to the same blocks as the blocks in the conventional image coding apparatus shown in FIG.

The image quality determination unit 160 includes a two-dimensional DCT coefficient of the original image input from the two-dimensional DCT conversion unit 151, a quantized coefficient of the two-dimensional DCT coefficient input from the linear quantization unit 152, and a quantization condition determination unit. An evaluation function value is calculated for each block from the quantization condition (quantization matrix) used at the time of quantization input from 180. And
Based on the calculation result, the degree of image quality deterioration of the image to be restored is determined using the quantization coefficient generated by the linear quantization unit 152, and if the degree of image quality deterioration is less than or equal to a predetermined standard, variable length coding is performed. For the unit 154, the linear quantization unit 15
2 is instructed to perform variable length coding on the quantized coefficient generated. On the other hand, the image quality determination unit 160, if the degree of image quality deterioration exceeds a predetermined standard, the quantization condition determination unit 180.
, And instruct to perform more accurate quantization.

The quantization condition determination unit 180 stores a predetermined quantization condition (quantization matrix) VTH0, and linearly quantizes one block of the two-dimensional DCT coefficient with respect to the linear quantization unit 152. As the quantization matrix for, the quantization condition VTH0 is first output. Then, thereafter, in accordance with an instruction from the image quality determination unit 160, the image quality determination unit 160 determines a quantization condition (quantization matrix) for performing finer quantization, and the value of the evaluation function value satisfies a predetermined condition. Continue to output until.

Next, the operation of the embodiment having the above structure will be described with reference to the flowchart of FIG. First, two-dimensional D
The CT conversion unit 151 divides the image signal 150 of the original image into blocks (for example, 8 × 8 pixels) composed of a plurality of pixels, and obtains the gradation values of the plurality of pixels in each block (8 × 8 pixels). (Gradation level value) is subjected to two-dimensional discrete cosine transformation to obtain the same number of two-dimensional DCT coefficients as the plurality of pixels,
The plurality of two-dimensional DCT coefficients of each block are output to the linear quantization unit 152 and stored in the memory in the image quality determination unit 160 (S1).

Next, the quantization condition determining unit 180 selects and outputs the quantization condition (quantization matrix) VTH0, which is predetermined and stored by itself, to the linear quantization unit 152 (S2).

The linear quantizer 152 uses the selected quantizing condition VTH0 to select the two-dimensional DCT transformer 151.
When a plurality of two-dimensional DCT coefficients of each block input from are quantized and all the two-dimensional DCT coefficients in each block are quantized, the quantized coefficients of all the two-dimensional DCT coefficients in each of these blocks are quantized. Is output to the image quality determination unit 160 (S3).

The image quality determination unit 160 receives the quantization coefficient (quantization matrix) used for performing the quantization processing in the processing S3 from the quantization condition determination unit 180 every time the quantization coefficient of each block is input. ) Is input to each block by using the quantized coefficient of each block, the quantization condition, and the two-dimensional DCT coefficient before quantization of each block stored in its own memory in the process S1. The evaluation function value of is calculated (S4).

The image quality judging section 160 uses, for example, an expression for evaluating an error due to quantization as an evaluation function for obtaining the evaluation function value. That is, as an example thereof, the evaluation function E that calculates the sum of squares of the quantization error (rate to the amplitude width) of the AC component for each block is used.

The formula of this evaluation function is shown below.

[0064]

[Equation 1]

(However, X = Y = 0, that is, DC component F
₀ (0, 0) is excluded) _{where, G 1 (X, Y)} = F 0 (X, Y) · Q (X, Y) ······ above formula, in, F ₀ (X, Y) is a two-dimensional DCT
Coefficient, F ₁ (X, Y) is the two-dimensional DCT coefficient F
It is a quantized coefficient obtained by quantizing ₀ (X, Y) with a quantization threshold Q (X, Y).

In the above equation, its denominator, that is, [F ₀ (X, Y) -G ₁ (X, Y)] is equal to the quantization error. The judgment of the image quality deterioration using the evaluation function shown in this equation is performed as follows. When a large image quality deterioration occurs, a large error occurs due to quantization, so that the denominator of the expression of the evaluation function becomes large and the evaluation function E becomes a small value. On the other hand, when the image quality deterioration is small, the evaluation function E has a large value because the error due to the quantization is small. Therefore, the value E of this evaluation function can be used to estimate the degree of image quality deterioration due to quantization.

After calculating the value of the evaluation function E, the image quality determination unit 160 determines whether the value of the evaluation function E is equal to or more than the allowable value (S5). Then, when the value of the evaluation function E is equal to or more than the allowable value, the image quality determination unit 160 uses the code table 155 in the variable length coding unit 154 to directly code the quantized coefficient in the variable length coding. (S7).

On the other hand, when the image quality judging unit 160 judges in the judging process S5 that the value of the evaluation function E is smaller than the allowable value, the quantizing condition deciding unit 180 is made to perform finer quantization. After instructing and changing the quantization condition (quantization matrix) by the quantization condition determination unit 180 (S7), the linear quantization unit 152 uses the changed quantization condition (quantization matrix), 2 above
Requantize the dimensional DCT coefficient F ₀ (X, Y) (S
3).

Hereinafter, in the determination processing S5, the quantization condition is changed and the value of the evaluation function E is calculated until the image quality determination unit 160 determines that the value of the evaluation function E is equal to or larger than the allowable value. When the image quality determination unit 160 determines that the value of the evaluation function E is equal to or greater than the allowable value (S5) (S3 to S6), the image quality determination unit 160 instructs the variable length encoding unit 154 to perform encoding to change the variable length. The encoding unit 154 inputs the quantization coefficient of each block from the linear quantization unit 152, and performs variable length encoding of the quantization coefficient of each block (S7).

Therefore, according to this embodiment, the two-dimensional DCT coefficient obtained by the two-dimensional DCT transform, the quantized coefficient obtained by quantizing the two-dimensional DCT coefficient, and the quantization used for the quantization. By determining the quantization condition (quantization matrix) using
It is possible to accurately control the image quality. In other words, it is possible to perform fine quantization processing for an image with many gradation change portions and coarse quantization for an image with many gradation change portions. It is possible to secure an image quality of a predetermined level or higher.

As the evaluation function for evaluating the deterioration degree of the image quality, for example, the following expressions using absolute values may be adopted in addition to those shown in the above expressions.

[0072]

[Equation 2]

Here, G ₁ (X, Y) = F ₁ (X, Y) · Q (X, Y) FIG. 7 shows the image quality judgment unit 160 in the embodiment shown in FIG.
It is a figure which shows one structural example. In the figure, the coefficient storage unit A165 stores the two-dimensional DCT coefficient for each block output from the two-dimensional DCT conversion unit 151 shown in FIG.

Further, the coefficient storage unit B166 stores the quantized coefficient output from the linear quantization unit 152. The address generation unit 164 uses the two coefficient storage units 165, 16 described above.
6 to the corresponding two-dimensional DCT in the same block
An address signal for synchronously reading the coefficient and its quantized coefficient is used as the two coefficient storage units 165, 16.
Output to 6.

The function calculation section 168 uses the coefficient storage section 16
5, 166 corresponding two-dimensional DC in the same block
The T coefficient F ₀ (X, Y) and the quantized coefficient F ₁ (X, Y) are input and stored in the coefficient storage unit B from the quantization condition determination unit 180 in synchronization with the input of these coefficients. Input the quantization threshold value Q (X, Y) used to obtain the quantization coefficient F ₁ (X, Y).

The function calculation unit 168 calculates the value E of the evaluation function of the AC component for each block by the processing of the above formula, and outputs the value E to the comparison unit 170. The upper limit storage unit 167 stores an allowable value δ of the value E of the evaluation function (hereinafter, referred to as an upper limit δ). The quantization condition storage unit 169 also stores the quantization condition (quantization matrix) used by the linear quantization unit 152 to obtain the quantization coefficient.

The comparing section 170 compares the value E of the evaluation function input from the function calculating section 168 with the upper limit value δ stored in the upper limit storage section 167, and the value E of the evaluation function is the upper limit. When the value is δ or more, the value in the allowable limit block number storage unit 171 is incremented. The allowable limit block number storage unit 171 stores the number of blocks in which the value E of the evaluation function is equal to or more than the upper limit value δ, that is, the image quality of the restored image exceeds the target.

The block number threshold value storage unit 172 includes the determination unit 1
73 stores the threshold value of the number of blocks for determining whether or not requantization is necessary. The determination unit 173 compares the number of blocks output from the allowable limit block number storage unit 171 with the block number threshold value stored in the block number condition storage unit 172, and determines that the value E of the evaluation function is the upper limit value δ or more. If there are more blocks than the block number threshold value, the variable length coding unit 154 is instructed to perform variable length coding on the quantized coefficient obtained by the linear quantization unit 152. In response to this instruction, the variable length coding unit 154 zigzag scans the quantization coefficient of the linear quantization unit 152,
Variable length coding is executed with reference to the code table 155, and the created code data is output. Then, the encoding process ends when the encoding of all the quantized coefficients of all the blocks of the original image is completed.

On the other hand, when the number E of the evaluation function value E is equal to or larger than the upper limit value δ is smaller than the block number threshold value, the determination unit 173 performs finer quantization on the quantization condition determination unit 180. The linear quantizing unit 152 is instructed to output the quantizing condition. Quantization condition determination unit 18
8 shows an example of the configuration of 0.

In the figure, the selection unit 181 receives three types of quantization conditions (quantization matrices) according to an instruction from the determination unit 173 in the image quality determination unit 160, and the linear quantization unit 152.
Selectively output to.

Quantization condition 1 storage unit 182, quantization condition 2
The storage unit 183 and the quantization condition 3 storage unit 183 store the quantization condition 1 (quantization matrix 1), the quantization condition 2 (quantization matrix 2), and the quantization condition 3 (quantization matrix 3), respectively. To do.

The relationship between these three kinds of quantization condition 1, quantization condition 2, and quantization condition 3 with respect to the deterioration of the image quality of the restored image is as follows:
The restored image is quantized in the order of quantization condition 1, quantization condition 2, and quantization condition 3 so that the restored image becomes finer.

That is, quantization condition 1, quantization condition 2,
In the order of the quantization condition 3, the quantization threshold for the DC component and each AC component of the two-dimensional DCT coefficient becomes smaller. Next, the operation of the image data encoding device shown in FIG. 5 having the image quality determination unit 160 and the quantization condition determination unit 180 having the above configuration will be described with reference to the flowchart of FIG.

First, the two-dimensional DCT converter 151 sequentially inputs the image signal 150 of a multivalued original image of one frame in block units of 8 × 8 pixels, and the blocks are divided into 8 × 8 2 blocks.
Converted to a block (two-dimensional DCT coefficient block) composed of three-dimensional DCT coefficients (SA1), and the 8 × 8 two-dimensional D
Store CT coefficient block in internal memory (SA
2). The two-dimensional DCT conversion unit 151 uses the series of processing S described above.
A1 and SA2 are performed for all blocks of the multivalued original image (SA1 to SA3).

Subsequently, the two-dimensional DCT transform section 151 converts the first two-dimensional DCT coefficient block of the multivalued original image into the one shown in FIG.
It is stored in the coefficient storage unit A165 in the image quality determination unit 160 shown in (1) and is also output to the linear quantization unit 152.

The image quality determination unit 160 determines the first two-dimensional D
Upon receiving the input of the CT coefficient block, the quantization condition determining unit 1
80 is instructed to select the quantization condition 1 as the quantization matrix. In response to this instruction, the quantizing condition determining unit 180 reads the quantizing condition 1 from the quantizing condition 1 storing unit 182 by the selecting unit 181 and selectively outputs it to the linear quantizing unit 152. The data is stored in the conversion condition storage unit 169 (SA4).

The linear quantization unit 152 uses the above quantization condition 1
Is input, each two-dimensional D in the first two-dimensional DCT coefficient block input from the above two-dimensional DCT transform unit 151.
The CT coefficient is linearly quantized using the quantization condition 1, and a block composed of 8 × 8 quantized coefficients obtained by the linear quantization is used as a coefficient storage unit B16 in the image quality determination unit 180.
6 (SA5).

Subsequently, the function calculating section 168 in the image quality determining section 160 has 63 two-dimensional DCT coefficients F ₀ (X, Y) in the first two-dimensional DCT coefficient block stored in the coefficient storage section A 165. , 63 quantization coefficients F ₁ (X, Y) in the coefficient storage unit B 166, and 63 quantization thresholds Q (X, X, of the quantization condition 1 stored in the quantization condition storage unit 169).
Y) is used to calculate the value E of the evaluation function represented by the above expression or, and the value E is output to the comparison unit 170 (S
A6).

The above variables (X, Y) are X = 0 to 7,
It takes a value of Y = 0 to 7. However, X = Y = 0 is excluded. Next, the comparison unit 170 compares the input evaluation function value E with the upper limit value δ stored in the upper limit value storage unit 167, and determines whether the evaluation function value E is greater than or equal to the upper limit value δ. That is, it is determined whether or not the error indicated by the value E of the evaluation function is within the allowable range (SA7), and if it is within the allowable range, the value of the allowable limit block number storage unit 171 is increased by "1" (increment). Yes (SA8). The allowable limit block number storage unit 171 is initialized to "0" before the processing shown in this flowchart is started.

The above series of processes SA5 to SA8 are performed for all the two-dimensional DCT coefficient blocks of the original image of one frame converted by the two-dimensional DCT converter 151 (SA5 to SA9).

When the processes SA5 to SA8 are completed for all the blocks of the two-dimensional DCT coefficient block (SA9), the determination unit 173 in the image quality determination unit 160 is performed.
Compares the effective block number stored in the allowable limit block number storage unit 171 with the block number threshold value stored in the block number threshold value storage unit 172, and if the effective block number is larger than the block number threshold value, (SA10, YE
S) Immediately, it instructs the variable length coding unit 154 to code the quantized coefficient of each block of the original image of one frame generated by the linear quantization unit 152. Upon receipt of this instruction, the variable length coding unit 154 codes the quantized coefficient of each block by the ADCT method (SA1).
2).

On the other hand, the determination unit 173 determines that the process SA10
If the number of effective blocks is equal to or less than the threshold value of the number of blocks, the quantization condition determination unit 180 is provided with a quantization condition for performing quantization such that the original image of one frame is more finely restored. The linear quantizing unit 152 is instructed to output. In response to this instruction, the quantization condition determination unit 180 causes the selection unit 181 to store the quantization condition 2 storage unit 183.
Quantization condition 2 for more accurate quantization from
Is output to the linear quantization unit 152 and the image quality determination unit 1
It is stored in the quantization condition storage unit 169 of 60 (SA1
1).

Then, using the quantization condition 2,
The above processing SA5 to SA10 is performed again, and D of all blocks is performed.
The linear quantization unit 152 performs linear quantization on the CT coefficient, and the function calculation unit 168 in the function value comparison unit 160 calculates the value E of the evaluation function.

Then, as described above, the determination unit 173 in the image quality determination unit 160 is again stored in the valid block number and block number threshold storage unit 172 stored in the allowable limit block number storage unit 171. When it is determined that the number of effective blocks is larger than the block number threshold value, that is, the quantization coefficient obtained by this linear quantization can restore the original image to a sufficient image quality (S
A10, YES), and the process SA12 causes the variable length coding unit 154 to code the original image using the quantized coefficient.

On the other hand, if the original image cannot be restored to a sufficient image quality even if the linear quantization is performed using the quantization condition 2, the determination unit 173 gives the quantization condition determination unit 180 more accuracy. Output to the quantization condition determination unit 180 an instruction to output a quantization condition for performing high quantization.
Upon receiving this instruction, the quantization condition determination unit 180 receives the instruction from the selection unit 1.
After 81, the quantization condition 3 is read from the quantization condition 3 storage unit 184, selectively output to the linear quantization unit 152, and stored in the quantization condition storage unit 169 in the image quality determination unit 160 (SA11). Returning to step SA5, the third quantization using the quantization condition 3 is started.

Then, the processes SA5 to SA9 are performed for all the blocks of the two-dimensional DCT coefficient block, and when the number of effective blocks is larger than the block number threshold value (SA10, YES), the variable length coding unit 154 determines that. Three
Variable length coding is performed on the quantized coefficient obtained by the quantized time (SA12).

As described above, three kinds of quantization conditions 1, 2,
3 is used to sequentially quantize the two-dimensional DCT coefficient block of the original image so that the quantization error becomes smaller. Then, when the quantized coefficients that can restore the original image to the target image quality are obtained, those quantized coefficients are variable length coded.

Contrary to the above embodiment, the number of blocks in which the original image cannot be restored to a sufficient image quality is counted, and if the number of blocks is a predetermined number or more, the quantization condition is changed and the quantization coefficient is changed. The blocks may be requantized.

Next, another configuration example of the quantization condition determining unit 180 is shown in FIGS. 10 (a), (b), (c) and (d). Each of these quantization condition determination units 180 has a basic condition storage unit 191 that stores one type of basic quantization condition (hereinafter referred to as a basic quantization condition), and the basic condition storage unit 191. A plurality of quantization conditions are generated and output using the basic quantization conditions stored in.

First, the quantization condition determining unit 1 shown in FIG.
The 80A is a linear condition by dividing a basic condition storage unit 191 that stores one type of basic quantization condition that is a basis for linearly quantizing a two-dimensional DCT coefficient block and each value of the basic quantization condition. The division unit 190 includes a division unit 190 that generates a plurality of quantization conditions used by the quantization unit 152, and a divisor storage unit 192 that stores a plurality of divisors used when the division unit 190 performs the division. In the quantization condition determination unit 180 having such a configuration, the initial value of the divisor storage unit 192 is, for example, "1", and the determination unit 173 in the image quality determination unit 160 receives the quantization condition selection signal. Each time, it is incremented by "1". Ie 2
At the time of the quantization process of the second time, the divisor stored in the divisor storage unit 192 becomes “2”, and the value of each matrix of the quantization conditions used by the linear quantization unit 152 for the second linear quantization is the first time. It becomes 1/2 of the value of. After that, in the same way, 3
The value of each matrix of the quantization conditions used for the linear quantization of the two-dimensional DCT coefficient block after the first time is 1 / sequentially.
It becomes 3, 1/4, ... Therefore, as shown in FIG.
It has the same function as the quantization condition determination unit 180 shown in FIG.

Further, the quantization condition determining unit 1 shown in FIG.
80b utilizes the characteristic of the two-dimensional DCT coefficient that the value of the high-order component becomes larger than the value of the low-order component,
As a divisor storage unit, a low-order divisor storage unit 192A that stores a divisor for a low-order two-dimensional DCT coefficient and a high-order two-dimensional DC
It has two high-order divisor storage units 192B for storing divisors for T coefficients. Low order coefficient storage unit 192A,
The high-order coefficient storage unit 192B stores, for example,
“2” and “3” are stored in advance, and every time the selection signal of the quantization condition is input from the determination unit 173 of the image quality determination unit 160, the low-order coefficient storage unit 192A and the high-order coefficient storage unit 19 are stored.
By appropriately changing the value stored in 2B and changing the divisor value for the low-order part and the divisor value for the high-order part at different rates, each two-dimensional DCT coefficient block is optimized for the optimum quantization condition (quantum Quantization matrix). It should be noted that the multiplexer (MPX) 199 sets the basic quantization condition stored in the basic condition storage unit 191 and the low quantization coefficient when the low-order quantized coefficient is generated according to the selection signal added from the linear quantization unit 152. The divisor stored in the next divisor storage unit 192A is selectively output, and when the next high-order quantized coefficient is generated, the basic quantization condition stored in the basic condition storage unit 191 The divisor stored in the next divisor storage unit 192B is selectively output to the division unit 192A.

Next, the quantization condition determining unit 1 shown in FIG.
80C has a divisor matrix storage unit 192C that stores the divisor values used by the division unit 100 to calculate the quantized coefficient in the zigzag scan order. That is, in the divisor matrix storage unit 192C, 8 × 8 individual division values corresponding to the quantization thresholds of the basic quantization conditions stored in the basic condition storage unit 191 one-to-one are zigzag-scanned. Stored in order. And the division unit 1
90 reads the quantization threshold value and the division value from the basic condition storage unit 191 and the divisor matrix storage unit 192C respectively in the zigzag scanning order, divides the quantization threshold value by the division value, and the value obtained by the division is quantized. The quantization threshold value of the quantization condition is output to the linear quantization unit 152.

Finally, in addition to the division unit 190 and the basic condition storage unit 191, the quantization condition determination unit 180D shown in FIG. 9D is the divisor used by the division unit 190 to generate the quantization threshold. A function storage unit 192D for generating is provided. The function storage unit 192D calculates a function y having x as a variable such as y = ax ² + bx + c, and outputs the calculation result to the division unit 190 as a divisor value.

This function storage unit 192D is used by the image quality determination unit 1
Each time the selection signal of the quantization condition is input from the determination unit 173 of 60, the value of the variable x is sequentially increased, the calculation of the function y is performed, and the division value having a larger value is sequentially output to the division unit 190. .. The division unit 190 divides each quantization threshold value of the basic quantization condition stored in the basic condition storage unit 190 by the division value, so that the linear quantization unit 152 quantizes the two-dimensional DCT coefficient block. Each quantization threshold value of the quantization condition at that time is sequentially changed to a small value in the order of the first, second, third, ... Quantization, and is output to the linear quantization unit 152.

The values stored in the high-order divisor storage unit 192A and the low-order divisor storage unit 192B described above are merely examples, and the values stored in these storage units are limited to these values. It is not something that will be done. Further, the function used by the function storage unit 192D for calculating the divided value is also the above-mentioned function y.
However, other various functions may be used.

Next, FIG. 11 is a diagram showing another example of the configuration of the image quality judging section. In the figure, blocks having the same functions as the blocks in the image quality determination unit 160 shown in FIG. 7 described above have the same names and reference numerals.

The image quality judgment unit 160A obtains the value of the evaluation function as the average value of the entire image. The function calculation unit 168A determines each block from the two-dimensional DCT coefficient and the quantization coefficient value stored in the coefficient storage units A165 and B166, and the quantization threshold of the quantization condition stored in the quantization condition storage unit 169. It has a function of accumulating the error of the quantized coefficient with respect to the original two-dimensional DCT coefficient of each degree for each. Then, the function calculation unit 168A accumulates the errors of the quantized coefficients of all blocks with respect to the original two-dimensional DCT coefficients, divides the accumulated value by the total number of blocks, and evaluates the entire image obtained by the division. Comparing the average of the function values 17
Then, the comparison unit 170 compares the average value with the upper limit value δ stored in the upper limit value storage unit 167. Other processes are the same as those of the image determination unit 160 shown in FIG.
Since the image quality determination unit 160a does not compare the conditions for each block, there is an advantage that the processing is simpler than that of the image determination unit 160.

FIG. 12 shows the image quality judgment unit 160A having the above structure.
6 is a flowchart for explaining the operation of the image data encoding device shown in FIG. First, the two-dimensional DCT conversion unit 151 performs the process S shown in the flowchart of FIG. 9 described above.
The same processing SB1 to SB3 as A1 to SA3 is performed to generate the two-dimensional DCT coefficient block for all blocks of the original image.

Subsequently, the image quality judgment unit 160 instructs the quantization condition determination unit 180 to cause the quantization condition determination unit 180 to output the first quantization condition to the linear quantization unit 152, and also quantize itself. The quantization condition is stored in the condition storage unit 169 (SB4).

Then, the linear quantizing unit 152 uses the two-dimensional D
Each time a two-dimensional DCT coefficient block is sequentially input from the CT transformation unit 151 and each two-dimensional DCT coefficient block is linearly quantized and converted into a quantized coefficient block using the input quantization condition (quantization matrix). Then, the quantized coefficient block obtained by the linear quantization is assigned to the image quality determination unit 1
It stores in the coefficient storage part 169 in 60A (SB5).

On the other hand, the function calculating section 1 of the image quality judging section 160A
68A linearly quantizes the two-dimensional DCT coefficient block stored in the coefficient storage unit A165 by the two-dimensional DCT conversion unit 151 and the two-dimensional DCT coefficient block stored in the coefficient storage unit B166 by the linear quantization unit 152. The value of the evaluation function is calculated from the quantized coefficient block obtained by the above and the quantization condition (quantization matrix) stored in the quantization condition storage unit 169 by the quantization condition determination unit 180 (SB6).

The processings SB5 and SB6 are repeatedly performed each time the quantized coefficient block is generated by the linear quantization unit 152, and the function calculation unit 168A accumulates (accumulates) the values of the evaluation function of each quantized coefficient block. ) Will continue. When the function calculating unit 168A finishes accumulating the evaluation function values of all the quantized coefficient blocks, the function calculation unit 168A divides the accumulated value by the number of blocks of all the quantized coefficient blocks to obtain the in-image average value of the evaluation function. To calculate.

Subsequently, the comparison section 170 compares whether or not the in-image average value of the evaluation function is larger than the upper limit value δ stored in the upper limit value storage section 167 (SB7), and the evaluation function If the average value in the image is larger than the upper limit value δ (SB
7, YES), and instructs the variable length coding unit 154 to perform coding, and in response to this, the variable length coding unit 154 sequentially inputs all the quantized coefficient blocks from the linear quantization unit 152 in a predetermined order. Then, the ADCT code data is generated and output (SB8).

On the other hand, if the in-image average value of the evaluation function is equal to or less than the upper limit value δ in the processing SB7, the comparing section 170 instructs the quantization condition determining section 180 to perform quantization for finer quantization. The condition is selected and output to the linear quantization unit 152, and the quantization condition is also stored in the quantization condition storage unit 169 (SB8).

Then, the above processing SB5 to SB8 is performed thereafter.
However, it is repeatedly performed until it is determined in the determination process SB7 that the in-image average value of the evaluation function is larger than the upper limit value δ. By the way, the function calculation unit 168A shown in FIG.
Every time the evaluation function value of each block is calculated, the evaluation function value is compared with the evaluation function value of the previous block calculated last time, and the evaluation function value with the smaller value is retained. The minimum value of the evaluation function value may be obtained at the stage of obtaining the evaluation function value for all the blocks, and the requantization may be performed when the minimum value is smaller than the upper limit value δ.

By using such a method, it is possible to judge whether or not requantization for reducing the quantization error is necessary with reference to the block in which the deterioration of image quality is most noticeable in the restored image.

The evaluation function is not limited to the above example, but the power (power spectral density) due to quantization is used.
It may be a function using the error of. Further, although the quantization condition is changed by division in the above embodiment, various operations other than division may be used.

Further, in the above embodiment, the orthogonal transform of the image data is performed by the discrete cosine transform (DCT), but the present invention is the discrete sine transform (DST), the discrete Legendre transform, and the Hadamard transform ( Hadamar
d), Harr transform, Karhunen-Loeve transform,
It is also applicable to an image data encoding device that encodes image data using other orthogonal transformation such as slant transformation.

[0119]

As described above, in the present invention, the degree of deterioration of the restored image is predicted, and the variable length coding is performed only when the degree of deterioration is within a predetermined allowable range. Therefore, it is possible to always encode the image data capable of restoring the high image quality regardless of the difference in the gradation change characteristics of the image.

[Brief description of drawings]

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

FIG. 2 is a block diagram (No. 2) of the principle of the present invention.

FIG. 3 is a principle block diagram (No. 3) of the present invention.

FIG. 4 is a principle block diagram (No. 4) of the present invention.

FIG. 5 is a system block diagram of an image data encoding device according to an embodiment of the present invention.

FIG. 6 is a flowchart illustrating the operation of the above-described embodiment.

FIG. 7 is a diagram illustrating a configuration example of an image quality determination unit.

FIG. 8 is a diagram illustrating a configuration example of a quantization condition determination unit.

9 is a flowchart illustrating an operation of the image data encoding device illustrated in FIG. 5 including the quantization condition determination unit illustrated in FIG.

FIG. 10 is a diagram showing four types of configuration examples of a quantization condition determining unit using calculation.

FIG. 11 is a diagram illustrating another configuration example of an image quality determination unit.

12 is a flowchart illustrating an operation of the image data encoding device shown in FIG. 5 having the image quality determination unit shown in FIG.

FIG. 13 is a basic block diagram of a conventional image data encoding device.

FIG. 14 is a diagram showing an example of one block of an original image signal.

FIG. 15: Two-dimensional DC obtained by two-dimensional DCT transformation
It is a figure which shows an example of a T coefficient block.

FIG. 16 is a diagram showing a configuration example of a quantization matrix.

FIG. 17 is a diagram showing an example of a quantized coefficient block.

FIG. 18 is a diagram showing a scanning order of quantized coefficients.

[Explanation of symbols]

 1, 11, 21, 31 Quantization condition output means 2, 12, 22, 32 Quantization means 3, 13, 23, 33 Coding means 4, 16, 25, 35 Coding judgment means 14 Image quality judgment means 15 Counting means 24 Evaluation Average Value Calculation Means 34 Evaluation Function Calculation Means

Claims (4)

[Claims] 1. The image data of an original image is divided into a plurality of blocks, the gradation values of a plurality of pixels in the blocks are subjected to orthogonal transformation for each of the plurality of divided blocks, and then, , An image data encoding device for generating code data for image restoration after quantizing a plurality of orthogonal transform coefficients of each block obtained by the orthogonal transform, in which a plurality of orthogonal transform coefficients in each block are quantized. Quantizing condition output means (1) for outputting a plurality of quantizing conditions used for quantization, and using the quantizing condition from the quantizing condition outputting means (1),
For each block, a quantizing means (2) for quantizing a plurality of orthogonal transform coefficients in each block, and an encoding means (3 for coding the quantizing coefficient generated by the quantizing means (2). ) And using the quantized coefficient and the quantization condition used by the quantization means (2) for the quantization, the plurality of orthogonal transform coefficients in each block are restored and the restored From the orthogonal transformation coefficients and the plurality of orthogonal transformation coefficients in each block, a predetermined evaluation function for evaluating the deterioration degree of the image of each block is calculated, and the value of the evaluation function obtained by the calculation. It is determined whether or not the degree of deterioration of the restored image of each block is within the allowable range, and if the degree of deterioration of the restored image of each block exceeds the allowable range, the quantization condition output means (1 ), The quantization error was If the degree of deterioration of the restored image of each block is within the permissible range, an instruction to output a quantization condition that is smaller than the quantization is given, and the encoding means (3) is instructed to perform the encoding. An image data encoding device comprising: an encoding determination means (4) for 2. The image data of an original image is divided into a plurality of blocks, the gradation values of a plurality of pixels in the blocks are subjected to orthogonal transformation for each of the plurality of divided blocks, and then, , An image data coding apparatus for generating coded data for image restoration after quantizing a plurality of orthogonal transform coefficients of each block obtained by the orthogonal transform, wherein a plurality of orthogonal transform coefficients in each block are quantized. Using a quantization condition output means (11) for outputting a plurality of quantization conditions used for the above and a quantization condition from the quantization condition output means (11), The quantization coefficient (12) for quantizing a plurality of orthogonal transformation coefficients and the coding means (13) for encoding the quantized coefficient generated by the quantization means (12) are provided. And above The substituting means (12) restores the plurality of orthogonal transform coefficients in each block using the quantization condition used for the quantization, and restores the restored orthogonal transform coefficients and the plurality of orthogonal transform coefficients in each block. Whether or not a predetermined evaluation function for evaluating the degree of deterioration of the image of each block is calculated from the orthogonal transformation coefficient and whether the value of the evaluation function obtained by the calculation satisfies a predetermined condition. An image quality determination means (14) for determining the number of blocks, and a counting means (15) for counting the number of blocks in the image data of the original image that satisfy the above-mentioned predetermined conditions based on the determination result of the image quality determination means (14) When the number of blocks counted by the counting means (15) is less than a predetermined number, the quantization condition output means (1
If the value of the evaluation function satisfies the predetermined condition, the encoding means (13) is instructed to 1) to output the quantization condition such that the quantization error becomes smaller than the previous quantization. ), A coding judgment means (1
6) An image data encoding device comprising: 3. The image data of an original image is divided into a plurality of blocks, the gradation values of a plurality of pixels in the blocks are subjected to orthogonal transformation for each of the plurality of divided blocks, and then, , An image data coding apparatus for generating coded data for image restoration after quantizing a plurality of orthogonal transform coefficients of each block obtained by the orthogonal transform, wherein a plurality of orthogonal transform coefficients in each block are quantized. Using a quantization condition output means (21) for outputting a plurality of quantization conditions used for the above and a quantization condition from the quantization condition output means (21), The quantization coefficient (22) for quantizing a plurality of orthogonal transformation coefficients and the coding means (23) for encoding the quantized coefficient generated by the quantization means (22) are provided. And above The substituting means (22) restores the plurality of orthogonal transform coefficients in each block using the quantization condition used for the quantization, and restores the restored orthogonal transform coefficients and the plurality of orthogonal transform coefficients in the block. From the orthogonal transformation coefficient, by calculating a predetermined evaluation function for evaluating the deterioration degree of the image of each block, the evaluation function values of all blocks are obtained, and the evaluation function values of all the blocks An evaluation average value calculating means (24) for obtaining an average value of evaluation function values of one block, and a quantization condition output means (21) based on the average value obtained by the evaluation average value calculating means (24). On the other hand, a coding determination means (25) which gives either one of a quantization condition output instruction for which a quantization error is smaller than that of the previous quantization or an encoding instruction to the encoding means (23). And Image data encoding apparatus characterized by a. 4. The image data of an original image is divided into a plurality of blocks, the gradation values of a plurality of pixels in the blocks are subjected to orthogonal transformation for each of the plurality of divided blocks, and then, , An image data coding apparatus for generating coded data for image restoration after quantizing a plurality of orthogonal transform coefficients of each block obtained by the orthogonal transform, wherein a plurality of orthogonal transform coefficients in each block are quantized. For each block, the quantization condition output means (31) for outputting a plurality of quantization conditions to be used for outputting and the quantization conditions from the quantization condition output means (31) The quantizing means (32) for quantizing a plurality of orthogonal transform coefficients, and the coding means (33) for coding the quantized coefficients generated by the quantizing means (32), and the quantizing means (32 Using the quantization condition used for the quantization, restores the plurality of orthogonal transform coefficients in each block, and from the restored orthogonal transform coefficient and the plurality of orthogonal transform coefficients in each block. , Calculating a predetermined evaluation function for evaluating the degree of deterioration of the image of each block,
Evaluation function calculation means (34) for obtaining an evaluation function value indicating that the quantization error is the largest among the evaluation function values obtained by the calculation, and the evaluation obtained by the evaluation function calculation means (34). Based on the function value, the quantization condition output means (31) is instructed to output the quantization condition such that the quantization error is smaller than the previous quantization, or the encoding means (33).
An image data encoding device, comprising: an encoding determination means (35) for instructing either one of the encoding instructions for