JP3306201B2 - Data compression device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION The present invention relates to a method for recording a color still image on a recording medium in accordance with the JPEG algorithm, for example.
The present invention also relates to a data compression device provided in a device for reproducing from the recording medium.

[0002]

2. Description of the Related Art Since image data has a large amount of information, direct processing requires an enormous memory capacity and increases the time required for data communication. Therefore, conventionally, image data is processed after being compressed by encoding.

One of such image data compression algorithms is JPEG (Joint Photographic Expert).
Group). This J
The algorithm (JPEG algorithm) recommended by PEG is composed of a plurality of processes such as a baseline process.
This is a basic function that encodes an 8-bit precision original image using discrete cosine transform (DCT transform) and Huffman coding, and can transmit, receive, and reproduce image information.

In the baseline process, the pixel data of each block constituting one screen is subjected to DCT conversion,
A T coefficient is obtained, and the DCT coefficient is quantized by a quantization table. That is, the DCT coefficient is quantized by being divided by each of the quantization coefficients constituting the quantization table. This quantized D
The CT coefficients are Huffman-coded and recorded on a recording medium, for example, as compressed image data.

[0005]

That is, when the quantization coefficient constituting the quantization table is changed, the DCT coefficient changes, thereby changing the compressed image data and affecting the image quality of the reproduced image. By utilizing such a phenomenon, a plurality of quantization tables are prepared, and the quantization tables are selectively used according to the properties of the image to be processed, so that a reproduced image with good image quality can be obtained. However, if the DCT coefficient data changes for each quantization table in such a configuration, the image data length after Huffman coding changes,
A problem arises in that the control of the data compression processing circuit for performing Huffman coding or the like becomes complicated.

An object of the present invention is to solve such a problem, and to provide a data compression apparatus which can improve the quality of a reproduced image and does not complicate the control of a data compression processing circuit. It is an object.

[0007]

A data compression apparatus according to the present invention quantizes input data using a first quantization table including a plurality of quantization coefficients, and performs a first quantization including a plurality of coefficient values. First quantizing means for obtaining data;
A second quantization unit that quantizes the input data by using a second quantization table including a quantization coefficient different from the quantization coefficient of the quantization table and obtains second quantization data including a plurality of coefficient values; A grouping means for grouping each coefficient value of the first and second quantized data in accordance with the data length, and a group to which the coefficient value of the second quantized data belongs to the first quantized data. In order not to be a group with a data length larger than the group to which the coefficient value of the data belongs,
Means for modifying the second quantized data , wherein the second quantity
Dequantization of multiple coefficient values using the quantization table
The sum of the squares of the error between the inverse quantization coefficient and the input data
Inverse quantization of multiple coefficient values using one quantization table
Sum of the error between the inverse quantization coefficient obtained by
So that the second quantization table is
It is characterized by what is required .

[0008]

BRIEF DESCRIPTION OF THE DRAWINGS FIG. FIG. 1 is a block diagram showing a still video camera provided with an image compression apparatus to which one embodiment of the present invention is applied.

A still image signal of the subject S obtained by the image pickup unit 10 is transmitted to an image compression device 20, subjected to data compression processing as described later, and recorded on a recording medium M such as an IC card. In the recording medium M, in addition to the image signal subjected to the data compression, a quantization table used when reproducing the image signal is recorded.

In the image pickup section 10, the light arriving from the subject S is condensed by a condensing lens 11, and a subject image is formed on a light receiving surface of a solid-state image pickup device (CCD) 12. C
A large number of photoelectric conversion elements are arranged on the light receiving surface of the CD 12,
On the upper surface of the photoelectric conversion element, for example, a color filter in which R, G, and B color filter elements are arranged in a predetermined manner is provided, and each photoelectric conversion element corresponds to one pixel data. . The subject image is converted into an electric signal corresponding to a predetermined color by each photoelectric conversion element, and is converted into an A / D signal.
It is input to the converter 13.

The signal digitally converted by the A / D converter 13 is converted into a luminance signal Y and color difference signals Cb and Cr by a signal processing circuit (not shown).
Is input to The image memory 14 is divided into mutually independent memory areas for storing the luminance signal Y and the color difference signals Cb and Cr, respectively, and each memory area has a storage capacity for one screen. The luminance signal Y and the color difference signals Cb and Cr once stored in the image memory 14 are
For data compression processing, first, the DCT of the image compression device 20 is used.
It is input to the processing circuit 21.

Although the DCT processing circuit 21 is shown as one processing circuit in FIG.
An independent DCT processing circuit is provided for each of the color difference signals Cb and Cr. In FIG. 1, a signal line extending from the imaging unit 10 to the image compression device 20 and a signal line connecting each component (such as the DCT 21) in the image compression device 20 are:
It is provided for transferring data corresponding to the luminance signal Y, and a signal line for transferring data relating to the color difference signals Cb and Cr is omitted.

In the image compression device 20, image data such as a luminance signal Y is processed in block units. As shown in FIG. 2, each pixel block R1 is composed of 8 × 8 pixel data, and this pixel data indicates, for example, a luminance signal Y. When one screen is composed of, for example, 768 × 480 pixels, the total number of blocks is 96 × 60 = 5.
760 blocks.

In the DCT processing circuit 21, the pixel data of the original image is two-dimensionally DCT-transformed for each block R1, thereby obtaining 8 × 8 DCT coefficients R2 for one block.
Is required. In a matrix of DCT coefficients R2,
The DCT coefficient data (260) in the upper left corner indicates a DC component, and the other DCT coefficient data indicates an AC component. Further, in this matrix, the components on the right side indicate components having a higher spatial frequency in the horizontal direction of the image, and the components on the lower side indicate components having a higher spatial frequency in the vertical direction of the image.

The DCT output from the DCT processing circuit 21
The coefficient R2 is quantized by the quantization processing circuit 22 using a quantization table (Q _Y , Q _Cb , Q _Cr ) R3. The quantization table R3 includes a luminance signal Y and color difference signals Cb and Cb.
Each component of the DCT coefficient R2 is divided by the corresponding quantization coefficient and quantized. That is, the number of components of the DCT coefficient R2 is the same as the number of quantization coefficients.
Each coefficient value of the quantized DCT coefficient (quantization table) R4 obtained by dividing the quantization table R3 by the DCT coefficient R2 is rounded off and converted to an integer. Some information has been lost.

The quantized DCT coefficient R4 is input to the inverse quantization processing circuit 24 and inversely quantized. That is, the quantized DCT coefficient R4 is multiplied by the quantization table (Q _Y , Q _Cb , Q _Cr ) R3 in the inverse quantization processing circuit 24, thereby obtaining the inverse quantized DCT coefficient R5 as shown in FIG. Can be Each component of the inversely quantized DCT coefficient R5 is exactly the same as the DCT coefficient R2 if no loss of image information occurs when the quantized DCT coefficient R4 is generated. Since a part of the information has been lost, it does not completely match the DCT coefficient R2, and the image quality of the reproduced image is somewhat deteriorated compared to the original image.
Therefore, in the present embodiment, as described below, the inverse quantization DC
The quality of a reproduced image is improved by minimizing the difference between the T coefficient R5 and the DCT coefficient R2, that is, the DCT quantization error.

The inversely quantized DCT coefficient R5 and the DCT coefficient R2 output from the DCT processing circuit 21 are input to a quantization error calculator 25, and the difference between these coefficients R5 and R2, that is, the DCT quantization error R6 is calculated. Is done. The optimum quantization coefficient calculation unit 26 calculates the optimum quantization table R7 using the DCT quantization error R6 as described below.

In the i-th block, the matrix of the quantization table R3 is Q _uv , the matrix of the DCT coefficient R2 is S _iuv , the matrix of the quantized DCT coefficient R4 is r _iuv , and the matrix of the inverse quantized DCT coefficient R5 is S '
_iuv . When the square sum E _uv of the DCT quantization error is obtained for one screen, E _uv = Σ (S ′ _iuv −S _iuv ) ² = Σ (r _iuv × Q _uv −S _iuv ) ² (1) Here, Σ indicates the total sum of all blocks constituting one screen.

[0019] (1) equation is differentiated for _{Q uv, dE uv / dQ uv} = 2Σr iuv - a _{(r iuv × Q uv S iuv} ) (2). Here, Q _uv when the sum of squares E _uv takes a minimum value
Is Q ′ _uv , this can be obtained by changing equation (2) to 0 and modifying the equation. That is, Q ′ _uv = round ( _{Σr iuv} · S _iuv / _{Σr iuv} ² ) (3) Here, round means rounding to an integer.

Optimum quantization table R7 according to equation (3)
Are calculated, and an optimum quantization table R7 is generated based on the calculation result. This optimal quantization table R7 is composed of quantization tables (Q _Y ′, Q _Cb ′, Q
_Cr ′) is recorded on the recording medium M. Note that, as understood from the comparison with the quantization table R3 in FIG. 2, the optimal quantization table R7 differs from the quantization table R3 in data surrounded by a circle.

The optimum quantization table R7 is input to the quantization processing circuit 22. That is, each component of the DCT coefficient R2 output from the DCT processing circuit 21 is divided and quantized in the quantization processing circuit 22 by the corresponding quantization coefficient of the optimal quantization table R7. Also, as shown in FIG. 4, each coefficient value of the optimum quantized DCT coefficient R10 obtained by this quantization is rounded off and converted to an integer as in the case of the quantized DCT coefficient R4.

The quantized DCT coefficient R4 and the optimal quantized DCT coefficient R10 obtained by the quantization processing circuit 22 are input to the group comparing section 31, and the respective coefficient values of the quantized DCT coefficient R4 and the optimal quantized DCT coefficient R10 Are grouped according to Table 1 as described later, and it is determined whether or not the corresponding coefficient values belong to the same group. According to this determination result, the optimal quantized DCT coefficient R10 is DC
The modified quantized DCT corrected by the T coefficient determination unit 32
The coefficient R13 is output to the Huffman encoding processing circuit 23.

[0023]

[Table 1]

The Huffman encoding processing circuit 23 determines a code word according to a predetermined encoding table based on a group number of each coefficient value and a run length (the number of times the coefficient value is 0), which will be described later. An additional bit corresponding to the group number is added to the word, whereby the DCT coefficient R13 is encoded into a series of binary data.

Next, the generation of the modified quantized DCT coefficient R13 by the group comparing section 31 and the DCT coefficient determining section 32 will be described with reference to FIGS. 4 and 5, and Tables 1 and 2.

Table 1 shows the grouping of the DC components of the quantized DCT coefficients R4 and R10 (the components located at the upper left corner of each matrix in FIG. 4, for example, "16" in the quantized DCT coefficient R4). ing. In Table 1, each coefficient value of the quantized DCT coefficient is grouped according to the data length (binary data) after encoding, and the data length of the DCT coefficient value increases as the group number increases. The minimum coefficient value is the DC value belonging to the group.
Among the T coefficient values, the one with the smallest absolute value and the maximum coefficient value indicate the one with the largest absolute value among the DCT coefficient values belonging to the group. Note that the group number of the AC component (the component located at a portion other than the upper left corner in FIG. 4) is from 0 to 10, the coefficient value of the group number 11 does not exist, and the other components are the same as the DC component. , Illustration is omitted.

[0027]

[Table 2]

Table 2 shows r _uv (quantization D) in the example of FIG.
CT coefficient R4), r ' _uv (optimal quantized DCT coefficient R1
0), group C _uv (first DCT coefficient group R1
1), C ' _uv (second DCT coefficient group R12) and r "
_{Among uv} (modified quantized DCT coefficients R13), those having a numerical value other than 0 are particularly shown. The sign of the maximum coefficient value is the same as the sign of _r'uv . The numbers enclosed in parentheses of the DC component (the uppermost data) correspond to difference values described later.

FIG. 5 shows the quantization DCT from the DCT coefficient R2.
A coefficient R4 and an optimal quantized DCT coefficient R10 are obtained, and a modified quantized coefficient R is calculated based on these DCT coefficients R4 and R10.
13 shows a flowchart of a program for calculating the number 13.

Step 101 is the same as the above-described quantization DCT.
This is a step of obtaining the coefficient R4 and the optimal quantized DCT coefficient R10, and this is executed in the quantization processing circuit 22. That is, in this step, r _uv (quantized DCT coefficient R4) is calculated from S _uv (DCT coefficient R2) using Q _uv (quantization table R3), and Q ′ _uv ( optimized quantization table R7) is calculated. Using S _uv (DCT coefficient R2)
r ′ _uv (the optimal quantized DCT coefficient R10) is calculated. When step 101 is executed once, r _uv and r ′ _{uv are calculated} for one block composed of 8 × 8 pixel data.
Is required.

In step 102, the group C is obtained from r _uv
uv (first DCT coefficient group R11) is calculated, and
r ′ _uv to group C ′ _uv (second DCT coefficient group R1
2) is calculated. The groups C _uv and C ′ _uv are determined by referring to Table 1 and detecting which group each coefficient value of r _uv and r ′ _uv belongs to. When step 102 is executed once, as in step 101, groups C _uv and C ′ _uv are obtained for each coefficient value for the quantized DCT coefficients r _uv and r ′ _uv of one block.

The method of grouping in step 102 differs between the DC component and the AC component, and the calculation of the groups C _uv and C ′ _uv relating to the DC component is performed based on the DC component of the block currently being processed and the previously processed DC component. The calculation is performed based on the difference value between the DC component of the block and the group number of the groups C _uv and C ′ _uv regarding the AC component, based on the coefficient value of the block currently being processed.

The calculation of the groups C _uv and C ′ _uv relating to the DC component will be described. For example, assuming that the value of the DC component of the previously processed block is “36”, since the coefficient value of r _uv is “16” in the example of FIG. 4, the difference value is “−20” (= 16). -36). Difference value "-20"
The group number of the group _Cuv to which is belongs is “5” with reference to Table 1. Similarly, since the coefficient value of _r'uv is "20", the difference value is "-16" (=
20-36), and the group number of the group C ′ _uv to which the difference value “−16” belongs is “5”.

The calculation of the group numbers of the groups C _uv and C ′ _uv relating to the AC component will be described. In the example of FIG. 4, r _uv
With respect to the coefficient value “4” on the right of the DC component of, the group number of the group C _uv is “3” with reference to Table 1.
For the coefficient value “7” on the right of the DC component of r ′ _uv , the group number of group C ′ _uv is “3”. Similarly, for the coefficient value “−7” on the lower side of the DC component of r _uv , referring to Table 1, the group number of the group C _uv is “3”, and the lower side of the DC component of r ′ _uv is “3”. The coefficient value "-
For “8”, the group number of group C ′ _uv is “4”.

When the groups C _uv and C ′ _uv are calculated for all coefficient values of one block in this way, step 10
In steps 3 and 104, parameters u and v are initialized to "0", and in steps 105 to 111, a modified quantized DCT coefficient r " _uv is obtained for one block. In the matrix, the address of each component of the row extending in the left-right direction is indicated, and increases from 0, 1, 2,..., 7 by 1 from left to right. Indicates the address of the component, and increases by one from 0 from top to bottom, that is, the parameters u and v of the DC component are each 0.

[0036] At step 105, the group C _'uv group number of whether greater than the group number of the group C _uv is determined. When C ′ _uv > C _uv, the maximum coefficient value is selected as the modified quantized DCT coefficient r ″ _uv in step 106. On the other hand, when C ′ _uv > C _uv is not _satisfied , in step 107, the modified quantum DCT coefficient r " _uv
r ' _uv is selected.

When step 105 is executed for the first time, the parameters u and v are each “0”, and the group number for the DC component is determined. In the case of FIG. 4 and Table 2, C ′ _uv = C _uv Therefore, step 107 is executed. That is, the optimal quantized DCT coefficient r ' _uv (20) is selected as the modified quantized DCT coefficient r " _uv . In other words, the data length of the difference value of the optimal quantized DCT coefficient r' _uv is the quantized DCT coefficient. If it is not longer than the data length of the difference value of r _uv , the optimal quantized DCT coefficient r ′ _uv is adopted.
If the data length of the difference value (coefficient value in the case of AC component) of the optimal quantized DCT coefficient _r'uv is longer than the data length of the difference value (coefficient value in the case of AC component) of the quantized DCT coefficient _ruv The maximum coefficient value is used in place of the optimal quantized DCT coefficient _r'uv , which will be described later.

In step 108, the parameter u is incremented by one, and in step 109, the parameter u is incremented.
Is determined to be 8 or more. Parameter u is 8
If smaller than DCT coefficient groups R11, R12
Since the processing for one row has not been completed in each matrix, the process returns to step 105 again.

At this time, assuming that the parameter u is 1 and the parameter v is 0, in the case of FIG. 4 and Table 2, r ′ _uv = 7, r _uv = 4, and C ′ _uv = C _uv ( = 3), step 107 is executed, and the modified quantized DC
The optimal quantized DCT coefficient r ' _uv (7) is selected as the T coefficient r " _uv . That is, the data length of the coefficient value of the optimal quantized DCT coefficient r' _uv is the data length of the coefficient value of the quantized DCT coefficient r _uv The optimal quantized DCT coefficient _r'uv is taken as the modified quantized DCT coefficient r " _uv .

On the other hand, in step 109, the parameter u
Is greater than or equal to 8, after the parameter v is incremented by 1 in step 110,
It is determined whether or not the parameter v is 8 or more. If the parameter v is smaller than 8, since processing has not been completed for all rows in each matrix of the DCT coefficient groups R11 and R12, step 104
, The parameter u is returned to zero, and the processing of steps 105 to 109 is performed for the next row.

At this time, for example, assuming that the parameter u is 0 and the parameter v is 1, in the case of FIG. 4 and Table 2, r ′ _uv = −8, r _uv = −7, and C ′ _uv > C _Since it is _uv , step 106 is executed, and the maximum coefficient value (= −7) is selected as the modified quantized DCT coefficient r ″ _uv . That is, the data length of the coefficient value of the optimal quantized DCT coefficient r ′ _uv is is longer than the data length of the coefficient values of the quantized DCT coefficients r _uv, the coefficient values of the group to the coefficient values of the quantized DCT coefficients r _uv (-7) belongs (-7, ..., 4,4, &・ ・ 、
Of the 7), the one (−7) closest to the coefficient value (−8) of the optimal quantized DCT coefficient r ′ _uv is the modified quantized DCT coefficient r ″
Adopted as _UV .

When the above operation is repeated and all the coefficient values of one block are set for the modified quantized DCT coefficient r " _uv , the parameter v is set to 8 in step 111.
It is determined that this is the case, and the program is once terminated. This program is executed for all blocks constituting one screen.

The modified quantized DCT coefficient r " _{uv for} one screen obtained according to the program shown in FIG. 5 is input to the Huffman encoding processing circuit 23 and encoded by a predetermined algorithm. Image signal (Y ', C
b ′, Cr ′), that is, the image signal subjected to the data compression processing is recorded on the recording medium M. Since Huffman coding is conventionally known, its description is omitted.

As described above, according to the present embodiment, the image signal (Y ', Cb', Cr ') coded based on the modified quantized DCT coefficient r " _uv is stored in the recording medium M in the optimum condition. Therefore, the quantization table R7 is recorded, so that the image signal (Y ′, Cb ′, Cr ′) recorded on the recording medium M is inversely quantized to D
When the CT coefficient is obtained, the DCT coefficient is
When the pixel data becomes very close to the T coefficient R2 and is calculated by the inverse DCT transform, the input pixel data (the pixel block R
1) is reproduced, and the image quality of the reproduced image is improved.

In this embodiment, the modified quantized DCT coefficient r "
the data length of each coefficient value of the _uv is configured not longer than the data length of coefficient values of the quantized DCT coefficients r _uv. Therefore, the data length processed in the Huffman coding processing circuit 23 is limited, and the control of the circuit 23 is simplified.

Next, another embodiment of the present invention will be described with reference to FIG. Table 3 shows the values of r _uv , r ' _uv , C _uv , C' _uv and r " _uv in the example of FIG. In addition to the numerical values, the minimum coefficient value is shown, and the sign of the minimum coefficient value is the same as the sign of _r'uv , like the maximum coefficient value.

[0047]

[Table 3]

FIG. 6 shows a flowchart of a program for obtaining the modified quantization coefficient R13 based on the DCT coefficients R4 and R10. In FIG. 6, steps corresponding to the steps in the flowchart of FIG. 5 are denoted by the same reference numerals, and only different processes will be described.

When it is determined in step 105 that C ′ _uv > C _uv , in step 121, C ′ _uv <C _uv
Is determined. When C ′ _uv <C _uv is not _satisfied , that is, when C ′ _uv = C _uv , r ′ _uv is selected as the modified quantized DCT coefficient r ″ _uv in step 107. On the other hand, C ′ _uv <C _{uv If uv} , in step 122 the minimum coefficient value is selected as the modified quantized DCT coefficient r " _uv .

For example, when the parameter u is 1 and the parameter v is 0, in the case of FIG. 4 and Table 3, r ′ _uv = 7 and r _uv =
4 and C ′ _uv = C _uv (= 3), so step 1
07 is performed, and the optimal quantized DCT coefficient r ′ _uv (7) is selected as the modified quantized DCT coefficient r ″ _uv . That is, the data length of the coefficient value of the optimal quantized DCT coefficient r ′ _uv is determined by the quantization D
If the data length is equal to the coefficient value of the CT coefficient r _uv , the optimal quantized DCT coefficient r ′ _uv is adopted as the modified quantized DCT coefficient r ″ _uv .

For example, when the parameter u is 0 and the parameter v is 3, in the case of FIG. 4 and Table 3, r ′ _uv = 0 and r _uv =
Since −1 and C ′ _uv = 0 and C _uv = 1, step 122 is executed, and the minimum coefficient value (−1) is selected as the modified quantized DCT coefficient r ″ _uv . Although, for example, if r ′ _uv = 3 and r _uv = 7, then C ′ _uv = 2 and C _uv = 3, so that in step 122 the modified quantized DCT coefficient r ″ _uv The minimum coefficient value (4) is selected. That is, the optimal quantized DCT coefficient r ′
If the data length of the coefficient values of the _uv is shorter than the data length of the coefficient values of the quantized DCT coefficients r _uv, the coefficient values of the group to the coefficient values of the quantized DCT coefficients r _uv (7) belongs (-7, ... ,
-4, 4,... 7), the one (4) closest to the coefficient value (3) of the optimally quantized DCT coefficient _r'uv is the modified quantization D
Adopted as CT coefficient r " _uv .

According to this embodiment, similarly to the embodiment of FIGS. 1 to 5, the effect of improving the image quality of the reproduced image can be obtained, but the data of each coefficient value of the modified quantized DCT coefficient r " _uv can be obtained. Since the length is configured to always be the same as the data length of the coefficient value of the quantized DCT coefficient r _uv , the data length processed in the Huffman encoding processing circuit 23 is always constant, and this circuit 23 Control becomes even easier.

In the above embodiment, the present invention is applied to an apparatus for compressing an image signal in accordance with the JPEG algorithm. However, the present invention can be applied to other data compression.

[0054]

As described above, according to the present invention, a data compression apparatus which can improve the quality of a reproduced image and does not complicate the control of a data compression processing circuit such as a Huffman coding processing circuit. Is obtained.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a still video camera including an image compression device to which an embodiment of the present invention has been applied.

FIG. 2 is a diagram illustrating an example of a pixel block, DCT coefficients, a quantization table, and quantized DCT coefficients.

FIG. 3 is a diagram illustrating examples of an inverse quantization DCT coefficient, a DCT quantization error, and an optimal quantization table.

FIG. 4 shows an optimal inverse quantized DCT coefficient, a first DCT coefficient group, a second DCT coefficient group, and a modified quantized DCT.
It is a figure showing an example of a coefficient.

FIG. 5 is a flowchart of a program for obtaining a modified quantization coefficient.

FIG. 6 is a flowchart of a program for obtaining a modified quantization coefficient according to another embodiment.

[Description of Signs] 10 imaging unit 20 image compression device M recording medium

Continuation of the front page (58) Field surveyed (Int.Cl. ⁷ , DB name) H04N 1/41-1/419 G06T 9/00 H03M 7/30 H04N 7/30

Claims (2)

(57) [Claims] A first quantization unit that quantizes input data using a first quantization table including a plurality of quantization coefficients to obtain first quantization data including a plurality of coefficient values; A second one comprising a quantized coefficient different from the quantized coefficient
Quantizing the input data using a quantization table of
Second quantizing means for obtaining second quantized data composed of a plurality of coefficient values, and grouping means for grouping each coefficient value of the first and second quantized data according to the data length The second quantized data so that the group to which the coefficient value of the second quantized data belongs does not become a group having a longer data length than the group to which the coefficient value of the first quantized data belongs. Means to correct
In addition, a plurality of coefficient values are inversely calculated using the second quantization table.
Error between the inverse quantization coefficient obtained by
The sum of squares of the difference is calculated using the first quantization table.
Inverse quantization coefficient obtained by inverse quantization of the coefficient value of
So that it is smaller than the sum of the squares of the error with the force data.
And a second quantization table is obtained . 2. The correction unit according to claim 1, wherein the group to which the coefficient value of the second quantized data belongs is a group having the same data length as the group to which the coefficient value of the first quantized data belongs. The data compression apparatus according to claim 1, wherein the second quantized data is modified.