JPH07115668A - Device and method for compressing image - Google Patents

Abstract

PURPOSE:To reduce the degradation of picture quality without changing the image compression ratio at the time of compressing image data for which the object of especially character images or the like is photographed from a short distance. CONSTITUTION:At the time of a short distance photographing mode, in initial image scanning, the total code amount Hc of quantized and encoded chrominance data is calculated in a chrominance code amount calculation circuit 40, the total code amount Hc of the chrominance data is subtracted 43 from a set total code amount Nt and the total code amount Hyt of luminance data is obtained. Then, in an alphat calculation circuit 44, a coefficient alphat corresponding to the total code amount Hyt of the luminance data is obtained and the coefficient alphat is supplied to a multiplier 33. In second picture scanning, by quantizing and encoding the luminance data by the value of a quantization table 32 multiplied by the coefficient alphat, more code amount in the set total code amount Nt is distributed to the luminance data than the chrominance data.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image compression apparatus and image compression method for compressing image data.

[0002]

2. Description of the Related Art Recently, along with the digitization of video equipment, the digitization of signal processing is also progressing in the field of cameras. Along with this, development of a system for recording a digital video signal as a digital signal on a recording medium has been actively conducted.

For example, an electronic still camera is one example. Electronic still cameras for analog recording on a floppy disk are already on the market, but various electronic still cameras for digital recording on a memory card have been developed. In this case, since the recording capacity of the memory card is limited, the image data compression technique is essential.

A known compression method for image data is DPCM (Defferencial Pulse Code Modulati).
on) or orthogonal transform (eg DCT; Discrete
Cosine Transform) and the like.

At present, there is a strong tendency to increase the compression rate, and it is said that the data compression method using DCT is advantageous in this limit. In addition, international standardization of the compression coding method for still images is JPEG (Joint Photographic Expert Grou), which is a subordinate organization of ISO / CCITT.
The baseline method that is used by p) and is the most used among them is also the compression coding method using DCT.

In addition, although the memory card has a limited recording capacity, the DCT image compression itself is variable-length coding, and therefore it is customary to control the code amount during image compression to guarantee the number of recorded images. Is.

FIG. 2 shows a DC used in an electronic still camera.
It is a block diagram which shows the example of T compression encoding system. As shown in the figure, this system comprises a memory controller 1, a frame memory 2, a DCT operation unit 3, an encoder 4, a card controller 5 and a memory card 6. A digital video signal is supplied to the memory controller 1 from the terminal 1a. This signal is an interlaced signal and is temporarily stored in the frame memory 2 for conversion into a non-interlaced signal. Then 8
The interlace signal is read out for each block by using × 8 pixels as one block, and is supplied to the DCT calculation unit 3.
The DCT operation unit 3 performs a discrete cosine transform on the block unit signal and supplies the transform coefficient to the encoder 4. The encoder 4 quantizes the supplied transform coefficient and then Huffman-encodes it to generate compressed data, and supplies the generated compressed data to the memory card 6 via the card controller 5.

FIG. 3 is a block diagram showing the configuration of the encoder 4. In the figure, reference numeral 11 is a terminal for inputting the conversion coefficient for each block from the DCT calculation unit 3. Reference numeral 12 is a quantization table that holds, for each block, a value for quantizing the transform coefficient for each block. Reference numeral 13 is a multiplier for multiplying the value of the quantization table 12 by a coefficient (scale factor) given from the multiplexer 21. Reference numeral 14 is a divider that divides the input conversion coefficient for each block by the value of the quantization table 12 after multiplication to perform quantization. 15
Is a coding unit for Huffman coding the quantized transform coefficient. 16, 17, 18, and 19 are encoding units 1 respectively.
4 is a code amount calculation circuit for accumulating the encoded data obtained in 4 for each group and further obtaining the result of multiplying the integrated value by the number of groups as the group total code amount. Reference numeral 20 is a coefficient αt corresponding to the set total code amount Nt based on the group total code amount obtained by the respective code amount calculation circuits 16, 17, 18, and 19.
Is an αt calculation circuit for calculating Reference numeral 21 is a multiplexer for switching the coefficient given to the multiplier 13. Reference numeral 22 is an output terminal for outputting encoded data.

Next, the operation of the encoder 4 will be described.
The encoder 4 scans one image twice, calculates the coefficient αt necessary for final encoding of the conversion coefficient of the image data in the first scan, and uses the coefficient αt in the next scan to calculate the conversion coefficient. Is quantized and encoded to output compressed data. The details of the operation will be described below.

In the first scan, first, each block of transform coefficients of image data is sequentially input from the terminal 11a over the entire image, and the divider 14 causes the quantizer table 12 to input this.
Quantization is performed by dividing by the value obtained by multiplying the value of by a coefficient.

Here, as shown in FIG. 4, the conversion coefficient of the image data forming one screen is divided into a plurality of blocks, and each block of the conversion coefficient is G1, G2, G3, G.
It is classified into four groups of four. The coefficients α1, α2, α3, α4 (α1 <α2 <
In the case of α3 <α4), the one associated with each group is selected by the multiplexer 21.

After that, the quantized transform coefficients are Huffman coded by the coding unit 14, and the coded data are integrated for each group by the code amount calculation circuits 16, 17, 18 and 19 of each group. Here, since the integrated value of each group is ¼ of the actual code amount, the integrated value of each group is multiplied by 4 and the result is the total code amount N1, N2, N for each group.
3, get as N4. Then, the total code amount N1, N
2, N3 and N4 are input to the αt calculation circuit 20.

By the way, in the encoder 4, for example, as shown in FIG. 5, a set total code amount Nt which influences the compression rate of image data and the image quality of compressed image is defined in advance.

The αt calculation circuit 20 includes the code amount calculation circuit 1
6, 17, 18 and 19, the total code amount N1, N2, N3, N4 for each group and the coefficients α1, α2, α3,
A coefficient αt corresponding to the set total code amount Nt is calculated from α4.

For example, if the set total code amount Nt is the value shown in the figure, A (N2, α2), B (N3, α) in the figure
The target coefficient αt can be determined from the linear equation approximated by the two points of 3).

Next, a second image scan is performed. In the second scan, the multiplexer 21 uses the αt calculation circuit 2
The coefficient αt obtained by 0 is selected and given to the multiplier 13. Then, the quantization table 1 multiplied by this coefficient αt
The transform coefficient of each block is quantized by the divider 14 using the value of 2. After that, the quantized transform coefficient of each block is Huffman coded by the coding unit 15, and is output from the output terminal 22 of the encoder 4 as compressed data.

The above-mentioned quantization and coding of image data are similarly performed for luminance data and color data.
That is, for both brightness and color, quantization is performed using the same quantization table value for each group.

By the way, such image compression fatally causes deterioration of image quality. This deterioration in image quality is due to block distortion caused by processing the image data block by block, ringing caused by loss of a wide area component, and the like. And this deterioration of image quality becomes a problem especially when photographing a subject with characters using an electronic still camera for professional use, for example, the periphery of the characters photographed is distorted and the characters cannot be distinguished. There is also.

[0019]

As described above, in the conventional image compression apparatus, the image quality is inevitably deteriorated, and when an image requiring a relatively high resolution such as a character image is compressed, the character cannot be discriminated. The problems such as are exposed.

The present invention is intended to solve such a problem, and particularly when compressing image data obtained by photographing a subject such as a character image from a short distance, deterioration of the image quality is reduced without changing the image compression rate. An object of the present invention is to provide an image compression apparatus and an image compression method that can be performed.

[0021]

In order to achieve the above-mentioned object, an image compression apparatus of the present invention divides luminance data and color data which are image data into a plurality of blocks, and transforms each block by orthogonal transformation. Orthogonal transformation means for obtaining coefficients, quantization means for quantizing the transformation coefficient of the luminance data and color data for each block obtained by the orthogonal transformation means using a quantization table, and the quantization means for quantizing Encoding means for encoding conversion coefficients of luminance data and color data, integrating means for integrating the total code amount of encoded data of color data obtained by the encoding means, and color data obtained by the integrating means. Depending on the subtraction means for subtracting the total code amount from the preset total code amount of the image data to obtain the distribution code amount to the luminance data, and the distribution code amount to the luminance data obtained by the subtracting means. The transform coefficients of the luminance data is provided with a table setting means for determining a value of the quantization table for quantizing.

[0022]

In the present invention, luminance data and color data, which are image data, are input twice to perform image compression. For the first input image, this is orthogonally transformed, quantized, and coded, then the total code amount of the coded data of the color data is calculated, and the calculated total code amount of the color data is set to a predetermined image. The distribution code amount to the brightness data is obtained by subtracting from the set total code amount of the data. Then, according to the distribution code amount to this luminance data,
The value of the quantization table for quantizing the conversion coefficient of the luminance data is determined. After that, the second image input is performed, the conversion coefficient of the luminance data is quantized and encoded using the value of the quantization table after the setting, and output as compressed data.
This makes it possible to allocate a larger amount of code to luminance data than to color data, as compared to a method in which both luminance data and color data are quantized / encoded using the same quantization table values. It is possible to improve the image quality of a short-distance shot image such as a character image without changing the.

[0023]

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

FIG. 1 is a block diagram showing the configuration of an encoder (quantization / encoding circuit) in an image compression apparatus according to an embodiment of the present invention.

In the figure, reference numeral 31 is a terminal for inputting the conversion coefficient of the image data for each block from the DCT calculation section. Reference numeral 32 is a quantization table holding a value for quantizing the transform coefficient for each block for each block. A multiplier 33 multiplies the value in the quantization table 32 by a coefficient (scale factor) selectively given from the second multiplexer 45. Reference numeral 34 is a divider that divides the input transform coefficient for each block by the value of the quantization table 32 after the coefficient multiplication to perform quantization. Reference numeral 35 is an encoding unit for Huffman encoding the quantized transform coefficient. 36, 37, 38,
Reference numeral 39 indicates the total code amount N1, N2, N3 for each group obtained by multiplying the coded data obtained by the coding unit 34 for each group and further multiplying the integrated value by the number of groups.
, N4 as four code amount calculation circuits. These code amount calculation circuits 36, 37, 38, 39 calculate only the total code amount of the luminance data of each group in the short-distance shooting mode, and in other cases, the total code of the luminance data and color data of each group. Works to calculate quantity. 40 integrates the coded data of the color data obtained by the coding unit 35,
A color code amount calculation circuit for obtaining the integrated value as a total code amount Hc of color data. Reference numeral 41 detects a distance between a subject (not shown) and an electronic still camera which is an image pickup means. 39 and the short distance detection circuit for outputting to the first and second multiplexers 42 and 45. 42 is a close-up photography mode selection signal S1
, The output of the color code amount calculation circuit 40 is selected,
In other cases, the first multiplexer selects "0" and supplies it to the subtractor 43. Reference numeral 43 denotes a subtracter that subtracts the value selected by the first multiplexer 42 from the set total code amount Nt set in advance for the entire image data (luminance data and color data). The total code amount Hc of the color data is subtracted from the code amount Nt and the result is obtained as the total code amount Hyt of the luminance data. Reference numeral 44 is an αt calculation circuit for calculating the coefficient αt used in the quantization of the image data input the second time. This αt calculation circuit 44
The output Nyt of the subtractor 43, the code amount calculation circuits 36 and 37,
38, 39 outputs N1, N2, N3, N4 and the coefficient α applied to the quantization of the image data in the first image scan
The coefficient αt is calculated using 1, α2, α3, and α4.
That is, the coefficient αt of the coefficient αt is based on the total code amount Hyt of the luminance data in the short-distance shooting mode, and based on the set total code amount Nt determined for the entire image data (luminance data and color data) at other times. Calculation is done. Note that the coefficient αt is calculated by α1, α as described in the conventional technique (FIG. 5).
2, α3, α4 and N1, N2, N3, N4 are linearly approximated from a table. 45 is a multiplier 3
3 is a second multiplexer that switches the coefficient to be given to 3; An output terminal 46 outputs the encoded data.

Next, the operation of this embodiment will be described. First, the short-distance detection circuit 41 detects the distance between the subject and the electronic still camera, and detects whether or not the shooting to be performed is short-distance shooting. Then, when it is detected that the short-distance photographing is performed, the short-distance detecting circuit 41 outputs the short-distance photographing mode selection signal S.
1 is output to each of the code amount calculation circuits 36, 37, 38, 39 and the first and second multiplexers 42, 45.

Next, the operation at the time of short-distance shooting will be described. In the first image scanning, first, each block of the conversion coefficient of the luminance data is sequentially input to the terminal 31 over the entire image, and this is divided by the divider 34 by the value obtained by multiplying the value of the quantization table 32 by the coefficient. And quantize. At this time, the coefficients are synchronized with the group of blocks input to the input terminal 31 from the second multiplexer 45 by α1, α2, α3, α4 (α1 <
α2 <α3 <α4) are respectively supplied.

After that, the conversion coefficient of the quantized luminance data is Huffman coded by the coding unit 35, and the coded data is coded by the code amount calculation circuits 36, 37, 38, 39 of each group.
Will be added for each group. Here, since the integrated value of each group is 1/4 of the actual code amount, the integrated value of each group is multiplied by 4 and the result is the code amount N1, N for each group.
2, N3, and N4 are input to the αt calculation circuit 44, respectively.

In the short distance shooting mode, the short distance detection circuit 4
By inputting the short-distance shooting mode selection signal S1 from 1 to each of the code amount calculation circuits 36, 37, 38, 39, the operation of each code amount calculation circuit 36, 37, 38, 39 is performed by the above-mentioned brightness data. Only code amount calculation ends.

Next, each block of color data conversion coefficients is sequentially input to the terminal 31 over the entire image, and this is divided by the divider 34.
Then, the value in the quantization table 32 is divided by a value obtained by multiplying the coefficient α4 to perform quantization. The coefficient α4 is the coefficient having the highest compression rate in the four groups.

After that, the conversion coefficient of the quantized color data is Huffman coded by the coding unit 35, the coded data is integrated by the color code amount calculation circuit 40, and the total of the color data as the integration result is obtained. The code amount Hc is input to the first multiplexer 42. Here, the first multiplexer 42 receives the short distance shooting mode selection signal S1 and receives the color code amount calculation circuit 4
The output Hc of 0 is selected and given to the subtractor 43. Then, the subtractor 43 subtracts the total code amount Hc of the color data from the set total code amount Nt, and inputs the result to the αt calculation circuit 44 as the total code amount Hyt of the luminance data.

The αt calculation circuit 44 calculates the total code amount Hyt of the input luminance data and the total code amount N of the luminance data of each group.
1, N2, N3, N4 and coefficients α1, α2, α3, α
Using 4 as a parameter, the coefficient αt used in the quantization of the luminance data input the second time is calculated. Here, the total code amount Hyt of the luminance data is a value obtained by subtracting the total code amount Hc of the color data from the set total code amount Nt, and the total code amount Hc of the color data is the coefficient α4 having the highest compression rate. Is a value derived from the result of quantization using. Therefore, as compared with the conventional method in which the same coefficient is used for both the luminance data and the color data, a larger code amount of the set total code amount Nt can be distributed to the luminance data. In other words, a part of the code amount of color data is allocated to the code amount of luminance data.

Next, the second image scan is performed.
At the time of quantizing the luminance data at this time, the coefficient αt calculated by the αt calculation circuit 44 is selected by the second multiplexer 45 and given to the multiplier 33. Thereafter, the luminance data is quantized and encoded in the same manner as described above, and the encoded data is output from the output terminal 46.

When the color data to be subsequently input is quantized, the coefficient α4 having the highest compression rate is selected by the second multiplexer 45 and given to the multiplier 33.
After that, the color data is quantized and encoded in the same manner as described above, and the encoded data is output from the output terminal 46.

Next, operations other than short-distance shooting will be described.
In this case, the same operation as the conventional one is performed. That is, since the short-distance shooting mode selection signal S1 is not input to the code amount calculation circuits 36, 37, 38, 39, the code amount calculation circuits 36, 37, 38, 39 respectively code the luminance data and the color data. The total code amount is calculated for the encoded data, and the results N1, N2, N3, and N4 are output to the αt calculation circuit 44.

At this time, the first multiplexer 42 selects a value of "0" and inputs it to the subtractor 43, so that the set total code amount Nt is directly input from the subtractor 43 to the αt calculation circuit 44. It will be.

As a result, the coefficient α commonly used in the quantization of the luminance data and the color data in the αt calculation circuit 44.
t is calculated, and the conversion coefficient of the luminance data and the color data input in the second image scan is quantized using this coefficient αt.

Thus, according to the image compression apparatus of the present embodiment, when performing short-distance shooting such as character image shooting, color data is quantized by using a coefficient having a high compression ratio and distributed to the color data. By suppressing the code amount as much as possible and increasing the code distribution amount to the luminance data by that amount, it is possible to improve the image quality without changing the compression rate.

The present invention is not limited to this embodiment, and various modifications can be made without departing from the spirit of the invention.

For example, in the above-described embodiment, the short-distance detection circuit 41 automatically detects short-distance photographing, but a changeover switch for selecting the short-distance photographing mode is provided to artificially change the short-distance photographing mode. It may be set.

In this embodiment, the image data is divided into four groups for controlling the code amount. However, the present invention is not limited to this, and the code amount can be controlled by the value of the quantization table. It is applicable in the range of the image compression device.

[0042]

As described above, according to the image compressing apparatus and the image compressing method of the present invention, a code to be distributed to color data when compressing image data obtained by photographing a subject such as a character image from a short distance. By suppressing the amount to be low and allocating a large amount of code to the luminance data by that amount, deterioration of the image quality can be significantly reduced without changing the image compression rate.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of an encoder (quantization / encoding circuit) in an image compression apparatus according to an embodiment of the present invention.

FIG. 2 is a block diagram showing an example of a DCT compression encoding system used in a conventional electronic still camera.

FIG. 3 is a block diagram showing a configuration of an encoder of the DCT compression encoding system in FIG.

FIG. 4 is a diagram showing grouping of image data.

FIG. 5 is a diagram showing a relationship between a coefficient given to a quantization table and a code amount.

[Explanation of symbols]

31 ... Input terminal, 32 ... Quantization table, 33 ... Multiplier, 34 ... Divider, 35 ... Encoding section, 36, 37, 3
8, 39 ... Code amount calculation circuit, 40 ... Color code amount calculation circuit,
41 ... Short range detection circuit, 42 ... First multiplexer, 4
3 ... Subtractor, 44 ... αt calculation circuit, 45 ... Second multiplexer, 46 ... Output terminal.

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI Technical display location H04N 1/41 C 7/30

Claims (3)

[Claims] 1. Brightness data and color data which are image data are divided into a plurality of blocks, and orthogonal transformation means for obtaining transformation coefficients by orthogonal transformation for each block, and luminance data for each block obtained by the orthogonal transformation means And a quantization means for quantizing the conversion coefficient of the color data using a quantization table, an encoding means for encoding the conversion coefficient of the luminance data and the color data quantized by the quantization means, and the encoding A totalizing means for integrating the total code amount of the coded data of the color data obtained by the means, and the total code amount of the color data obtained by the integrating means, by subtracting from the preset total code amount of the image data. Subtracting means for obtaining the distribution code amount to the luminance data, and determining the value of the quantization table for quantizing the conversion coefficient of the luminance data according to the distribution code amount to the luminance data obtained by the subtracting means. Image compression apparatus characterized by comprising a table setting means for. 2. The image compression apparatus according to claim 1, wherein the value of the quantization table for quantizing the conversion coefficient of the color data is the code of the color data quantized and coded using this quantization table. An image compression apparatus, wherein the amount is set to be smaller than a code amount obtained by quantizing and encoding the transform coefficient of the luminance data. 3. A step of inputting luminance data and color data, which are image data, a step of dividing the inputted luminance data and color data into a plurality of blocks, and orthogonal transforming each block to obtain a transform coefficient. The step of quantizing the conversion coefficient of the luminance data and the color data for each block using the quantization table, the step of encoding the quantized conversion coefficient of the luminance data and the color data, and the step of encoding the encoded data of the color data A step of calculating a total code amount, a step of subtracting the calculated total code amount of the color data from a preset total code amount of the image data to obtain a distributed code amount to the luminance data, and a step of calculating the luminance data Deciding the value of the quantization table for quantizing the conversion coefficient of the luminance data according to the distribution code amount of, and after inputting the input image data again after the value of the quantization table is decided. And force, image compression method characterized by having orthogonal transform every luminance data and color data, and obtaining a compressed data by performing quantization and coding.