JP2002016808A - Coding and decoding device, coding and decoding methods, and computer-readable record medium for executing methods - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a coding device for simplifying a coding circuit and simplifying quantization by utilizing the positional correlation of each component before color transformation, and then by carrying out the color transformation. SOLUTION: In each buffer shown in Figure 2, image data consisting of components of colors R, G, and B read from a scanner is subjected to transformation utilizing position correction for each color of R, G, and B in each of position correlation conversion parts 504, 505, and 506. After that, the DC constituent of each component is subjected to color transformation at a color conversion part 507 to become DC brightness and DC color signals, and then is quantified at quantization and coding parts 510 and 511. The AC component of each component is subjected to a first-stage quantization at an AC component quantization part 508, is subjected to the color transformation at a color transformation part 509 to become AC brightness and AC color signals, and is subjected to a second-stage quantization at the quantization and coding parts 510 and 511.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an encoding apparatus, a decoding apparatus, an encoding method, a decoding method and an encoding method for use in an image forming apparatus such as a digital copying machine and a facsimile apparatus. The present invention relates to a computer-readable recording medium.

[0002]

2. Description of the Related Art In recent years, demand for color images has been increasing year by year.
In addition, higher resolution is being developed in order to respond to demands for higher image quality. Under such circumstances, there is an increasing demand for compressing color images to save memory.

Conventionally, as a method of compressing a color image, a method of first performing color conversion to divide image information into a lightness signal and a color signal, and then performing quantization to reduce the color signal more than the lightness signal has been used. . This is based on the idea of efficiently leaving information of high importance by concentrating information on the brightness signal by color conversion.

[0004] For example, No. 1 described in the proceedings of the “14th Annual Meeting of the Institute of Image Electronics Engineers of Japan in 1986” held on June 3 and 4 of 1986. FIG. 40 shows a method in "Encoding a color image based on a uniform color space" (Oshima, Kadoma, Kubota). In FIG. 40, color image data that has been subjected to color conversion (RGB → Lab conversion) in the RGB-Lab conversion unit 101 are subjected to color difference averaging units 112 and 122 to produce color signals a and b that have little effect on image quality even if they are deteriorated.
Only four pixels are averaged in 2 × 2 pixel block units.
After that, the image data includes the brightness signal and the chrominance signal, respectively, of the luminance sampling unit 103, the color difference
3 and is encoded by the encoding unit 134.

FIG. 41 shows a technique based on a similar idea in the technique described in Japanese Patent Application Laid-Open No. 63-9282. FIG.
In the color image data color-converted by the color conversion unit 202 for each color (R, G, B) block unit, the brightness signal is orthogonally transformed by the luminance orthogonal transformation unit 203 and converted into a DC component and an AC component. The DC and AC components are both quantized and coded by the quantization and coding unit 204, while the color signal is represented by a representative color (one color representing the information in the block by the color difference representative color calculation units 213 and 223). After the block average value is calculated in the embodiment, the quantization / encoding units 214 and 224
, Only the representative color is quantized and encoded.

FIG. 42 shows a technique described in Japanese Patent Application Laid-Open No. 6-245079. In FIG. 42, the color conversion unit 3
02, the color image data color-converted for each color in a predetermined unit is converted into a luminance DC / AC
Both the lightness signal and the chrominance signal are converted into DC / AC components by an AC conversion unit 313, and the quantization / encoding units 304, 314, 32
In 4, the AC components of the brightness signal and the color signal are quantized using the ratio of their amplitudes, and the quantization / encoding unit 314,
At 324, the DC component of the color signal is quantized using non-linear quantization such that information with a lower color difference is finely quantized.

As described above, in the conventional color image coding, information is first biased by color conversion, and then, if necessary, the information is further biased using positional correlation for each of the brightness signal and the color signal. Was the basic processing flow.

[0008]

However, such a conventional method is not always the most suitable for encoding a color image, and is wasteful in terms of circuit scale. Specifically, by not using a positional correlation having a stronger correlation than the correlation between colors upstream of the processing, the number of color conversion circuits that should be small or small in number is increased, or color conversion is performed. Therefore, there is a problem that separate quantization tables must be provided for the lightness signal and the color signal by performing quantization from.

SUMMARY OF THE INVENTION The present invention has been made in view of the above-mentioned problem, and simplifies an encoding circuit and performs quantum conversion by first utilizing the positional correlation of each component before performing color conversion and then performing color conversion. It is an object of the present invention to provide an encoding device, a decoding device, an encoding method, a decoding method, and a computer-readable recording medium for executing the method, which can simplify the encoding.

[0010]

In order to solve the above-mentioned problems, the present invention according to claim 1 compresses a color image composed of a plurality of components, each of which is a color component constituting a color image, by compressing code information. And a first conversion unit that converts information of each component so as to cause a bias in the amount of information using one-dimensional or two-dimensional positional correlation of an image, and the first conversion unit. A second conversion means for converting the information after conversion by the first conversion means to obtain a lightness signal and a color signal, and a quantum conversion means for converting the coefficient obtained by the conversion of the first conversion means and / or the second conversion means. And quantization means for performing quantization.

According to the above invention, the first conversion means converts the information of each component so as to cause a bias in the information amount using the one-dimensional or two-dimensional positional correlation of the image, and the second conversion means. Performs conversion on the information after conversion by the first conversion unit to obtain a lightness signal and a color signal, and the quantization unit is obtained by conversion of the first conversion unit and / or the second conversion unit. Quantization is performed on the coefficients.

According to a second aspect of the present invention, there is provided an encoding apparatus which generates code information by compressing a color image composed of a plurality of components, each of which is a color component constituting a color image. Separating means for separating a DC component and an AC component for each predetermined unit, and converting means for converting the AC component separated by the separating means into a brightness signal and a color signal to obtain an AC brightness signal and an AC color signal And a quantizing means for quantizing the AC lightness signal and / or the AC color signal obtained by the converting means.

According to the above invention, the separating means separates the information of each component into a DC component and an AC component for each predetermined unit, and the converting means converts the AC component separated by the separating means into a brightness signal and a color signal. To obtain an AC lightness signal and an AC color signal, and the quantization means quantizes the AC lightness signal and / or the AC color signal obtained by the conversion means.

According to a third aspect of the present invention, there is provided an encoding apparatus which generates code information by compressing a color image composed of a plurality of components, each of which is a color component constituting a color image. Separation means for separating a DC component and an AC component for each predetermined unit, AC component quantization means for quantizing the AC component separated by the separation means, and a brightness signal for separating the DC component separated by the separation means A first conversion means for converting the AC component quantized by the first quantization means into a brightness signal and a color signal. A second conversion means for obtaining an AC lightness signal and an AC color signal; and a quantization means for quantizing the AC lightness signal and / or the AC color signal converted by the second conversion means. That.

According to the above invention, the separating means separates the information of each component into a DC component and an AC component for each predetermined unit, and the AC component quantizing means quantizes the AC component separated by the separating means. The first converting means converts the DC component separated by the separating means into a brightness signal and a chrominance signal to obtain a DC lightness signal and a DC chrominance signal, and the second converting means performs quantization by the first quantizing means. The converted AC component is converted into a brightness signal and a chrominance signal to obtain an AC brightness signal and an AC chrominance signal, and the quantizing unit converts the AC brightness signal and / or the AC chrominance signal converted by the second conversion unit. Is quantized.

According to a fourth aspect of the present invention, there is provided a color image encoding apparatus for compressing a color image composed of a plurality of components, each of which is a color component constituting a color image, wherein information of each component is stored in a predetermined unit. Separation means for separating a DC component and an AC component for each, a conversion means for converting the DC component separated by the separation means into a brightness signal and a color signal to obtain a DC brightness signal and a DC color signal, Quantization means for quantizing the AC component separated by the means. According to the invention, the separating unit separates the information of each component into a DC component and an AC component for each predetermined unit, and the converting unit converts the DC component separated by the separating unit into a brightness signal and a color signal. Then, the DC brightness signal and the DC color signal are obtained, and the quantization means quantizes the AC component separated by the separation means.

The invention according to claim 5 is based on claim 2.
Alternatively, in the invention described in claim 3, the quantization means performs vector quantization when quantizing the AC brightness signal and / or the AC chrominance signal. According to the invention, the quantization means performs vector quantization when quantizing the AC brightness signal and / or the AC chrominance signal.

The invention according to claim 6 is the invention according to claim 2, 3, or 5, wherein
The quantization means quantizes the ratio between the AC lightness signal and the AC color signal when quantizing the AC lightness signal and the AC color signal. According to the above invention, the quantization means includes:
When quantizing the AC lightness signal and the AC color signal, the ratio between the AC lightness signal and the AC color signal is quantized.

The invention according to claim 7 is based on claim 4.
In the invention described in (1), the quantization means vector-quantizes a specific color having the largest edge among the AC components, and quantizes a ratio of another color to a vector of the specific color. According to the above invention, the quantization means includes:
Of the AC components, vector quantization is performed on the specific color having the largest edge, and the ratio of the other colors to the vector of the specific color is quantized.

The invention according to claim 8 is the invention according to claim 2.
The encoding device according to any one of claims 1 to 7, wherein the quantization unit performs vector quantization.
The threshold is set to be larger than the intermediate point of the quantization value. According to the above invention, when performing the vector quantization, the quantization means sets the threshold value to be larger than the intermediate point of the quantization value.

According to a ninth aspect of the present invention, there is provided the third aspect of the present invention.
In the invention described in (1), the AC component quantization means includes:
Quantization is performed by deleting lower bits of one coefficient or all coefficients. According to the invention, the AC component quantization means performs quantization by deleting lower bits of one coefficient or all coefficients.

According to a tenth aspect of the present invention, in the third aspect of the invention, the AC component quantizing means comprises:
The coefficient corresponding to the diagonal edge of the image is deleted. According to the above invention, the AC component quantization means deletes the coefficient corresponding to the oblique edge of the image.

The invention according to claim 11 is the invention according to any one of claims 2 to 10, wherein
The separation means performs conversion with reference to only pixels inside a predetermined unit which is a conversion unit. According to the above invention, the separation unit performs the conversion with reference to only the pixels inside the predetermined unit which is the conversion unit.

According to a twelfth aspect of the present invention, in the eleventh aspect, the separation means performs a Haar wavelet transform. According to the above invention, the separating means performs the Haar wavelet transform.

According to a thirteenth aspect of the present invention, there is provided a color image coding apparatus for generating code information by compressing a color image composed of a plurality of components which are color components forming a color image. Conversion means for performing BTC conversion of the information of
The color conversion device includes a color conversion unit that performs color conversion on the TC-converted information, and a quantization unit that quantizes the information that has been color-converted by the color conversion unit. According to the above invention, the conversion unit performs BTC conversion on the information of each component for each predetermined unit, the color conversion unit performs color conversion on the BTC converted information, and the quantization unit performs the color conversion on the information converted by the color conversion unit. Is quantized.

According to a fourteenth aspect of the present invention, there is provided a color image encoding apparatus for compressing a color image composed of a plurality of components which are color components constituting a color image to generate code information. First conversion means for converting the information of the component using a one-dimensional or two-dimensional positional correlation of an image for each predetermined unit so that the amount of information is biased; and after conversion by the first conversion means. Second conversion means for performing conversion on the information of (i) and (ii) to obtain a lightness signal and a color signal; area determination means for determining an image area for each of the predetermined units; and first conversion means and / or Quantizing means for performing different quantization for each of the coefficients obtained by the conversion by the second converting means in accordance with the determination result of the area determining means; and determining the area determined by the area determining means. A flag information generation means for generating a flag information for another, but with the.

According to the invention, the first converting means converts the information of each component so that the information amount is biased by using the one-dimensional or two-dimensional positional correlation of the image for each predetermined unit. , The second converting means converts the information after the conversion by the first converting means, the second converting means obtains a lightness signal and a color signal, and the area judging means performs the conversion of the image every predetermined unit. The region is determined, and the quantization unit performs different quantization on the coefficient obtained by the conversion by the first conversion unit and / or the second conversion unit for each region in accordance with the determination result of the region determination unit, and outputs the flag information. The creating means creates flag information for identifying the area determined by the area determining means.

According to a fifteenth aspect of the present invention, in the coding apparatus for generating code information by compressing a color image composed of a plurality of components which are each color component constituting a color image, For separating a DC component and an AC component for each predetermined unit, a first quantizing unit for quantizing the AC component separated by the separating unit, and a quantizing unit for quantizing by the first quantizing unit. Conversion means for converting the AC component into a brightness signal and a color signal to obtain an AC brightness signal and an AC color signal; and for each of the predetermined units, the AC component, the AC brightness signal, and the AC color signal. Area determining means for determining which of the plurality of areas the predetermined unit belongs to according to the magnitude of at least one amplitude of the signal; Alternatively, a second quantizer for performing different quantization for each of the AC color signals in accordance with the determination result of the area determiner, and flag information for identifying the area determined by the area determiner are created. And flag information generating means.

According to the above invention, the separating means separates information of each component into a DC component and an AC component for each predetermined unit, and the first quantizing means quantizes the AC component separated by the separating means. The converting means converts the AC component quantized by the first quantizing means into a brightness signal and a color signal to obtain an AC brightness signal and an AC color signal, and the area determining means performs , The AC component, the AC lightness signal, and the AC color signal, according to at least one of the amplitudes, determines which of the plurality of regions the predetermined unit belongs to, and the second quantization means The brightness signal and / or the AC color signal are subjected to different quantization for each area according to the determination result of the area determining means, and the flag information creating means is configured to output flag information for identifying the area determined by the area determining means. To create.

The invention according to claim 16 is the invention according to claim 15, wherein the second quantizing means sets the AC component or the AC brightness signal in a region where the amplitude is smaller. And / or coarsely quantizes the AC color signal. According to the above invention, the second quantizing means sets the AC component,
Alternatively, the AC brightness signal and / or the AC color signal are roughly quantized.

According to a seventeenth aspect of the present invention, in the invention according to any one of the fourteenth to sixteenth aspects, the area determining means determines an edge area and a non-edge area. According to the above invention, the area determination means determines an edge area and a non-edge area.

The invention according to claim 18 is the invention according to any one of claims 14 to 17, wherein the area judging means sets the area of n stages (n is an integer of 3 or more). It is to judge. According to the above invention, the region determination means determines a region of n stages (n is an integer of 3 or more).

According to a nineteenth aspect of the present invention, in the invention according to any one of the fourteenth to eighteenth aspects, the flag information is a flag-only bit or integrated with a code representing a coefficient value. It was decided to be composed. According to the above invention, the flag information may be a bit dedicated to the flag, or may be configured integrally with a code representing the coefficient value.

According to a twentieth aspect of the present invention, there is provided the method according to any one of the first to nineteenth aspects, wherein
A fixed-length code is generated as the code information.
According to the above invention, a fixed-length code is generated as code information.

According to a twenty-first aspect of the present invention, there is provided a decoding apparatus for decoding code information of a color image composed of a plurality of components, each of which is a color component constituting a color image, wherein the code information is dequantized. Inverse quantization means for restoring an AC lightness signal, an AC color signal, a DC lightness signal, and a DC color signal, and a first inverse means for inversely transforming the AC lightness signal and the AC color signal to restore an AC signal. A color conversion unit, a second inverse color conversion unit for performing inverse color conversion of the DC lightness signal and the DC color signal to restore a DC signal, and an AC signal restored by the first inverse color conversion unit. AC inverse quantizing means for quantizing, and the DC signal restored by the second inverse color transforming means and the AC signal inversely quantized by the AC component inverse quantizing means are subjected to inverse positional correlation conversion to obtain the color signal. Each component of the image Reverse position correlation transform means for restoring the Nento, those having a.

According to the invention, the inverse quantization means inversely quantizes the code information to restore the AC lightness signal, the AC color signal, the DC lightness signal and the DC color signal, and the first inverse color conversion means A second inverse color conversion unit for performing inverse color conversion on the brightness signal and the AC color signal to restore the AC signal, and a second inverse color conversion unit for performing inverse color conversion on the DC brightness signal and the DC color signal to restore the DC signal; Is inversely quantized of the AC signal restored by the first inverse color conversion means, and the inverse positional correlation conversion means is the DC signal restored by the second inverse transformation means and the AC signal inversely quantized by the AC inverse quantization means. The signal is inversely correlated to recover each component of the color image.

According to a twenty-second aspect of the present invention, there is provided a decoding apparatus for decoding code information of a color image composed of a plurality of components which are color components forming a color image, wherein the code information is dequantized. An inverse quantization means for restoring the AC lightness signal, the AC color signal, and the DC signal, and a first inverse color conversion means for inversely transforming the AC lightness signal and the AC color signal to restore the AC signal; AC inverse quantization means for inversely quantizing the AC signal restored by the first inverse color conversion means, and inverse positional correlation transform of the DC signal and the AC signal inversely quantized by the AC inverse quantization means. Inverse position correlation conversion means for restoring each component of the color image.

According to the above invention, the inverse quantization means restores the AC lightness signal, the AC color signal, and the DC signal by inversely quantizing the code information, and the first inverse quantization means performs the AC lightness signal and the AC color signal. The AC signal is restored by inverse color conversion of the signal, the AC inverse quantization means inversely quantizes the AC signal restored by the first inverse color conversion means, and the inverse positional correlation conversion means restores the DC signal and AC inverse quantization. The AC signal dequantized by the means is subjected to inverse positional correlation conversion to restore each component of the color image.

According to a twenty-third aspect of the present invention, in a decoding apparatus for decoding code information of a color image composed of a plurality of components, each of which is a color component constituting a color image, the code information is dequantized. Inverse quantization means for performing inverse color conversion on the information inversely quantized by the inverse quantization means, and inverse BTC conversion on the information inversely color transformed by the inverse color conversion means to produce the color Inverse BTC means for restoring each component of the image. According to the above invention, the inverse quantization means inversely quantizes the code information, the inverse color conversion means performs inverse color conversion on the information inversely quantized by the inverse quantization means, and the inverse BTC means performs inverse color conversion on the inverse color conversion means. The color-converted information is subjected to inverse BTC conversion to restore each component of the color image.

According to a twenty-fourth aspect of the present invention, in a decoding apparatus for decoding code information of a color image composed of a plurality of components which are color components constituting a color image, the code information belongs to the flag information. A region determining unit for determining a region, and performing a different inverse quantization on the code information of the predetermined unit for each region in accordance with a determination result of the region determining unit to obtain a DC brightness signal and a DC color signal, and an AC brightness signal and an AC brightness signal. Inverse quantization means for restoring a color signal; first inverse color conversion means for inverse color transforming the AC lightness signal and the AC color signal to restore an AC signal; Second inverse color conversion means for performing inverse color conversion to restore a DC signal, AC inverse quantization means for inverse quantizing the AC signal restored by the first inverse color conversion means, Inverse position correlation conversion means for performing inverse position correlation conversion on the DC signal restored by the inverse color conversion means and the AC signal inversely quantized by the AC inverse quantization means to thereby restore each component of the color image; , Is provided.

According to the above invention, the area determination means determines the area to which the code information belongs from the flag information, and the inverse quantization means separates the code information of a predetermined unit for each area according to the determination result of the area determination means. To restore the DC lightness signal and the DC color signal, and the AC color signal and the AC color signal, and the first inverse color conversion means performs the inverse color conversion of the AC lightness signal and the AC color signal to convert the AC signal. The second inverse color conversion means performs inverse color conversion on the DC brightness signal and the DC color signal to restore the DC signal, and the AC inverse quantization means performs AC recovery on the AC color signal restored by the first inverse color conversion means. Dequantizes the signal,
The inverse position conversion means performs inverse position correlation conversion of the DC signal restored by the second inverse color conversion means and the AC signal inversely quantized by the AC inverse quantization means to restore each component of the color image.

According to a twenty-fifth aspect of the present invention, in a decoding apparatus for decoding code information of a color image composed of a plurality of components which are color components constituting a color image, the code information belongs from the flag information. Area determining means for determining an area, and performing a different inverse quantization on the code information of the predetermined unit for each area according to the determination result of the area determining means, to obtain a DC brightness signal, a DC color signal, an AC brightness signal, and an AC Inverse quantizing means for restoring a color signal, first inverse color transforming means for inversely transforming the AC lightness signal and the AC color signal to restore an AC signal, and DC restored by the inverse quantizing means. Inverse position conversion means for performing an inverse position correlation conversion on the signal and the AC signal inversely quantized by the first inverse color conversion means to restore each component of the color image. Than it is.

According to the above invention, the area determining means determines the area to which the code information belongs from the flag information, and the inverse quantization means separates the code information of a predetermined unit for each area according to the determination result of the area determining means. To restore the DC lightness signal and the DC color signal, and the AC lightness signal and the AC color signal, and the first inverse color conversion means performs the inverse color conversion of the AC lightness signal and the AC color signal to restore the AC signal. The inverse position transforming means inversely transforms the DC signal restored by the inverse quantizing means and the AC signal inversely quantized by the first inverse color converting means to restore each component of the color image. I do.

According to a twenty-sixth aspect of the present invention, in the invention according to any one of the twenty-first to twenty-fifth aspects, the inverse quantization means uses an intermediate value when the DC brightness signal is restored. It is to restore. According to the above invention, the inverse quantization means restores the DC brightness signal with an intermediate value.

According to a twenty-seventh aspect of the present invention, there is provided an encoding method for compressing a color image composed of a plurality of components, each of which is a color component constituting a color image, to generate code information. A first step of transforming the information using a one-dimensional or two-dimensional positional correlation of the image so as to cause a bias in the amount of information; and performing a conversion on the information converted in the first step to obtain a brightness signal A second step of obtaining a color signal; and a third step of performing quantization on the coefficient obtained by the conversion of the first step and / or the second step.

According to the above invention, the information of each component is converted using the one-dimensional or two-dimensional positional correlation of the image so that the amount of information is biased, and the converted information is converted to obtain a brightness signal. And color signals are obtained, and the obtained coefficients are quantized.

According to the twenty-eighth aspect of the present invention,
In an encoding method for compressing a color image composed of a plurality of components as color components constituting a color image to generate code information, the information of each component is converted into a DC component and an AC component for each predetermined unit. First to separate
A second step of converting the separated AC component into a brightness signal and a color signal to obtain an AC brightness signal and an AC color signal; and quantizing the AC brightness signal and / or the AC color signal. And a third step of forming According to the above invention, information of each component is separated into a DC component and an AC component for each predetermined unit, and the separated AC component is converted into a brightness signal and a color signal, and the AC brightness signal and the AC color signal are converted. Then, the AC lightness signal and / or the AC color signal are quantized.

According to a twenty-ninth aspect of the present invention, in the encoding method for generating code information by compressing a color image composed of a plurality of components which are color components forming a color image, A first step of separating the separated DC component into a DC component and an AC component for each predetermined unit, a second step of quantizing the separated AC component, and a step of dividing the separated DC component into a brightness signal and a color signal. A third step of obtaining a DC lightness signal and a DC color signal by converting
A fourth step of converting the quantized AC component into a brightness signal and a color signal to obtain an AC brightness signal and an AC color signal;
A fifth step of quantizing the converted AC lightness signal and / or AC color signal.

According to the above invention, information of each component is separated into a DC component and an AC component for each predetermined unit, the separated AC component is quantized, and the separated DC component is converted into a brightness signal and a color signal. To obtain a DC lightness signal and a DC color signal, and convert the quantized AC component into a lightness signal and a color signal to obtain an AC lightness signal and an AC color signal.
The converted AC lightness signal and / or AC color signal are quantized.

According to a thirtieth aspect of the present invention, there is provided an encoding method for compressing a color image composed of a plurality of components, each of which is a color component constituting a color image, to generate code information. A first step of separating the DC component into a DC component and an AC component for each predetermined unit, and a second step of converting the separated DC component into a brightness signal and a color signal to obtain a DC brightness signal and a DC color signal. Process and
And a third step of quantizing the separated AC signal.

According to the above invention, the information of each component is separated into a DC component and an AC component for each predetermined unit, and the separated DC component is converted into a brightness signal and a color signal to convert the DC brightness signal into a DC signal. A color signal is obtained and the separated AC signal is quantized.

According to a thirty-first aspect of the present invention, in the coding method for obtaining code information by compressing a color image composed of a plurality of components which are each color component forming a color image, The method includes a first step of performing BTC conversion for each predetermined unit, a second step of performing color conversion on the BTC-converted information, and a third step of quantizing the color-converted information. According to the invention, the information of each component is BTC-converted for each predetermined unit, and the BTC-converted information is color-converted,
The color-converted information is quantized.

According to a thirty-second aspect of the present invention, there is provided an encoding method for generating code information by compressing a color image composed of a plurality of components, each of which is a color component constituting a color image. A first step of converting the information into a bias using a one-dimensional or two-dimensional positional correlation of an image for each predetermined unit, and performing a conversion on the converted information to obtain a brightness signal A second step of obtaining a color signal; a third step of determining an image area for each of the predetermined units; and a coefficient obtained by the conversion of the second step and / or the third step. On the other hand,
The method includes a fourth step of performing different quantization for each area in accordance with a result of the area determination, and a fifth step of creating flag information for identifying the determined area.

According to the above invention, the information of each component is converted so as to cause a bias in the amount of information by using a one-dimensional or two-dimensional positional correlation of an image for each predetermined unit, and the converted information is converted into the converted information. A conversion is performed to obtain a lightness signal and a color signal, a region of the image is determined for each predetermined unit, and a different quantization is performed on the coefficient obtained by the conversion for each region in accordance with the region determination result. Step 4 and flag information for identifying the determined area are created.

According to a thirty-third aspect of the present invention, there is provided an encoding method for generating code information by compressing a color image composed of a plurality of components, each of which is a color component constituting a color image. A first step of separating the AC component into a DC component and an AC component for each predetermined unit, a second step of quantizing the separated AC component, and converting the quantized AC component into a brightness signal and a color signal. A third step of obtaining an AC lightness signal and an AC color signal by converting the signal into a signal, and for each of the predetermined units, an amplitude of at least one of the AC component, the AC lightness signal, and the AC color signal. A fourth step of determining which of the plurality of areas the predetermined unit belongs to according to the size; and for the AC lightness signal and / or the AC color signal,
The method includes a fifth step of performing different quantization for each area according to the determination result of the area, and a sixth step of creating flag information for identifying the determined area.

According to the above invention, information of each component is separated into a DC component and an AC component for each predetermined unit, the separated AC component is quantized, and the quantized AC component is converted into a brightness signal and a color signal. Into an AC signal, an AC color signal and an AC color signal, and for each predetermined unit, the AC component, the AC lightness signal, and the AC color signal according to the amplitude of at least one of the above-described predetermined values. It is determined which unit the unit belongs to among the plurality of regions, and the AC brightness signal and / or the AC color signal are subjected to different quantization for each region according to the region determination result, and the determined region is identified. Create flag information for

According to a thirty-fourth aspect of the present invention, in the decoding method for decoding code information of a color image composed of a plurality of components which are color components constituting a color image, the code information is dequantized. A first step of restoring an AC lightness signal, an AC color signal, a DC lightness signal, and a DC color signal, and a second step of performing an inverse color conversion of the AC lightness signal and the AC color signal to restore an AC signal A third step of performing inverse color conversion on the DC lightness signal and the DC color signal to restore a DC signal; and a fourth step of inversely quantizing the AC signal restored by the first inverse color conversion unit. And a fifth step of performing an inverse position correlation transformation on the restored DC signal and the inversely quantized AC signal to restore each component of the color image.

According to the above invention, the AC information signal, the AC color signal, the DC lightness signal, and the DC color signal are restored by dequantizing the code information, and the AC lightness signal and the AC color signal are inverse-color-converted to perform AC conversion. The signal is restored, the DC lightness signal and the DC color signal are inverse-color-converted to restore the DC signal, the restored AC signal is inversely quantized, and the restored DC signal and the inversely quantized AC signal are inverted. Correlation conversion is performed to restore each component of the color image.

According to a thirty-fifth aspect of the present invention, in the decoding method for decoding code information of a color image composed of a plurality of components which are color components constituting a color image, the code information is inversely quantized. A first step of restoring an AC lightness signal, an AC color signal, a DC lightness signal, and a DC color signal; and a second step of performing an inverse color conversion of the AC lightness signal and the AC color signal to restore an AC signal. A third step of performing inverse color conversion on the DC lightness signal and the DC color signal to restore a DC signal, a fourth step of inversely quantizing the restored AC signal, and the restored DC signal. And a fifth step of inversely correlating the inversely quantized AC signal to restore each component of the color image.
And the step of

According to the above invention, the AC information signal, the AC color signal, the DC lightness signal, and the DC color signal are restored by dequantizing the code information, and the AC lightness signal and the AC color signal are inverse-color-converted to perform AC conversion. The signal is restored, the DC lightness signal and the DC color signal are inverse-color-converted to restore the DC signal, the restored AC signal is inversely quantized, and the restored DC signal and the inversely quantized AC signal are inverted. The components of the color image are restored by correlation conversion.

According to a thirty-sixth aspect of the present invention, in the decoding method for decoding code information of a color image composed of a plurality of components which are color components constituting a color image, the code information is dequantized. A second step of performing inverse color conversion of the inversely quantized information, and an inverse BTC of the information inversely converted by the inverse color conversion means.
Converting and restoring each component of the color image. According to the above invention,
The code information is inversely quantized, the inversely quantized information is inversely color-transformed, and the information subjected to the inverse color conversion by the inverse color conversion means is inversely BTC-transformed to restore each component of the color image.

According to a thirty-seventh aspect of the present invention, in the decoding method for decoding code information of a color image composed of a plurality of components, each of which is a color component constituting a color image, the code information belongs to the flag information. A first step of judging an area, and performing a different inverse quantization on the code information of the predetermined unit for each area according to the judgment result of the area to obtain a DC lightness signal and a DC color signal, and an AC lightness signal and an AC color signal. A second step of restoring the AC lightness signal and the AC color signal, and a third step of restoring the AC signal by performing an inverse color conversion on the AC lightness signal and the AC color signal. A fourth step of restoring the DC signal, a fifth step of dequantizing the restored AC signal, and an inverse position phase transforming of the dequantized AC signal and the restored DC signal. Convert and is intended to include, a sixth step of restoring each component of the color image.

According to the above invention, the area to which the code information belongs is determined from the flag information, and the code information of the predetermined unit is subjected to different inverse quantization for each area in accordance with the determination result of the area, so that the DC lightness signal and the DC color The signal, the AC light signal, and the AC color signal are restored, the AC light signal and the AC color signal are inverse-color-converted to restore the AC signal, and the DC light signal and the DC color signal are inverse-color-converted to convert the DC signal. The restored AC signal is inversely quantized, and the inversely quantized AC signal and the restored DC signal are subjected to inverse positional correlation transformation to restore each component of the color image.

According to a thirty-eighth aspect of the present invention, in a decoding method for decoding code information of a color image composed of a plurality of components, each of which is a color component constituting a color image, the code information belongs to the flag information. A first step of determining an area, and a second step of performing a different inverse quantization on the code information of the predetermined unit for each area according to a result of the area determination to generate a DC signal, an AC color signal, and an AC color signal. A third step of performing an inverse color conversion of the AC lightness signal and the AC color signal to restore the AC signal; a fourth step of inversely quantizing the restored AC signal; and And a fifth step of performing inverse positional correlation conversion on the DC signal and the inversely quantized AC signal to restore each component of the color image.

According to the above invention, the area to which the code information belongs is determined from the flag information, and according to the result of the area determination,
A DC signal, an AC chrominance signal, and an AC chrominance signal are generated by performing a different inverse quantization on the code information of a predetermined unit for each region,
The AC lightness signal and the AC color signal are inverse-color-converted to restore the AC signal, the restored AC signal is inversely quantized, and the restored DC signal and the inversely quantized AC signal are inversely position-correlated transformed. Restore each component of the color image.

According to a thirty-ninth aspect of the present invention, there is recorded a program for executing each step of the invention according to any one of the twenty-sixth to thirty-eighth aspects. According to the above invention, the program recorded on the recording medium is executed by a computer to realize each step of the invention according to any one of claims 26 to 38.

[0067]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of the present invention will be described below in detail with reference to FIGS.

In the present specification, “converting into a lightness signal and a color signal” is simply referred to as “color conversion”, and therefore, not only when converting RGB colors, but also when converting a frequency conversion coefficient into a lightness signal and a color signal. The conversion using the same conversion is also referred to as “color conversion”, and a circuit that performs such conversion is referred to as a “color conversion unit” or a “color conversion circuit”. This is because an achromatic image is common to normal color conversion in that a signal corresponding to a color signal becomes 0.

In this specification, the term "quantization" refers to means for widely reducing information. For example, means for reducing the number of bits by simple bit reduction or table conversion, or when all information is discarded. Is also described as quantization. In this specification, HL, LH, and HH are collectively referred to as “H”.
For example, "YH" means YHL, YL
H, YHH.

FIG. 1 is a block diagram showing a configuration of a signal processing system of a digital color copying machine according to the present embodiment. Here, a case will be described in which a digital color copier encodes color image data read from a scanner and decodes encoded code information.

The digital color copying machine shown in FIG. 1 reads a color image data (R, G, B) of a document by a scanner 401 and encodes the color image data (R, G, B) inputted from the scanner 401. (Compressing) encoder 40
2, a page memory 403 for storing the compressed code (code information) encoded by the encoder 402, and a page memory 4
03, a decoder 404 for decoding the coded compression code (code information) stored in the storage unit 03, and filtering / color conversion (RGB → CMYK) on the color image data decoded by the decoder 404. A filter / gradation processing unit 405 that performs processing such as conversion) / gradation processing and outputs the result to the writing system
And

Next, the operation of the signal processing system of the digital copying machine will be described. R read from the scanner 401
Image data composed of components of three colors of GB is encoded (compressed) by an encoder 402, and stored in a page memory 403.
Is stored in In this embodiment, for the sake of convenience such as rotation and editing of an image, a fixed-length code system that allows easy access to a memory is adopted as an image encoding method. Here, the component refers to each color component forming a color image such as RGB or CMYK. The image data in the page memory is read out to the editing panel 406 and edited as needed. At the time of writing (at the time of image formation), the compressed code (code information) compressed by the encoder 402 and stored in the page memory 403 is read from the page memory 403 in units of 2 × 2 pixel blocks, and decoded. 404,
Filter processing by filter / gradation processing unit 405, RGB-
After being subjected to CMYK conversion, gradation processing, etc., it is transferred to the writing system.

Hereinafter, the encoder 402 and the decoder 4 shown in FIG.
The first to sixth embodiments of the fourth embodiment will be described in detail.

(Embodiment 1) FIG.
FIG. 3 is a block diagram showing Embodiment 1 of the present invention. The encoder 402 illustrated in FIG. 2 receives R, G, and B input from the scanner 401 in units of n × m (for example, in units of 2 × 2 pixel blocks).
R, G, and B color n × m buffers 501, 502, and 503 for temporarily storing image data composed of color components (for example, 8 bits for each pixel of each color);
Conversion using position correlation (hereinafter referred to as “position correlation conversion”) is performed for each component (RGB) stored in the n × m buffers 501, 502, and 503 of the B color to separate a DC component and an AC component for each color. Then, while outputting the DC component to the color conversion unit 507, the R, G, and B color position correlation conversion units 504 and 50 output the AC component to the AC component quantization unit 508.
5, 506 and R, G, B color position correlation conversion unit 50
A color conversion unit 507 that performs color conversion (brightness / color separation) on DC components of each component input from 4, 505, and 506 to generate a DC lightness signal and a DC color signal, and outputs the DC lightness signal and the DC color signal to the quantization unit 510; It has.

The encoder 402 shown in FIG.
Color, G, and B color position correlation conversion units 504, 505, and 506
The AC component quantization unit 508 that performs the first-stage quantization on the AC component of each component input from the AC component and outputs the result to the color conversion unit 509, and the first-stage quantization that is input from the AC component quantization unit 508. A color conversion unit 509 that performs color conversion (brightness / color separation) on the AC component of each of the quantized components to generate an AC brightness signal and an AC color signal, and outputs the generated signal to the quantization unit 510; The DC lightness signal and the DC color signal input from the color conversion unit 509 are quantized, and the AC lightness signal and the AC color signal input from the color conversion unit 509 are subjected to a second-stage quantization, and are encoded. 511, and information (2) quantized by the quantization unit 510.
And an encoding unit 511 that creates a fixed-length code of (× 2 pixel block units) and stores it in the page memory 403.

Next, an encoding method of the encoder 402 of FIG. 2 will be described. In FIG. 2, components of R, G, and B colors (for example, 2
× 2 pixel block units, each pixel has 8 bits per pixel), the R, G, and B color nxm buffers 501, 5
02, 503 and temporarily stored.

The R, G, and B color position correlation conversion unit 504,
Reference numerals 505 and 506 denote n × m buffers 50 of R, G and B colors.
1, 502, and 503 (R
GB) to perform a position correlation conversion to obtain a DC component L for each color.
LR, LLG, LLB and the AC components HR, HG, HB are separated, and the DC components LLR, LLG, LLB are separated by the color conversion unit 50.
7, while the AC components HR, HG, and HB are output to the AC component quantization unit 508.

Here, as the position correlation conversion, for example,
HarrWavele, a type of wavelet transform
A t-transform (hereinafter called "Haar transform") can be used in two dimensions. The two-dimensional Haar transform will be specifically described. The two-dimensional Haar transform is performed in units of 2 × 2 pixel blocks, and FIG. 3 shows an example of a 2 × 2 pixel block (R,
G, B). In the present embodiment, as shown in FIG. 3, after a pixel block is cut out for each of RGB, RGB
The Haar transform as shown in the equation of FIG. 4 is performed for each color. As a result of the Haar transform, the image information of the 2 × 2 pixel block is LL,
It is converted into four coefficients of HL, LH and HH. LL
The coefficients are DC components, and the HL, LH, and HH coefficients are AC components.

In the present embodiment, the case where the Haar transform is used as the position correlation transform (means for separating into a DC component and an AC component) has been described. However, the present invention is not limited to this. As a position correlation transform, a sub-band transform as a generic term of orthogonal transform such as DCT transform, Slant transform, Hadamard transform, and frequency transform, and a block code utilizing image density that has a moderate gradation change locally. And the like are also applicable.

However, as in the present embodiment, the code information is stored in an internal memory (page memory 40) of the digital color copying machine.
In the usage form stored in 3) and accessed a plurality of times, conversion in which conversion is performed with reference to information of adjacent pixel blocks, that is, overlap conversion is not preferable. That is, in a case where the code information stored in the memory is accessed a plurality of times to form a plurality of copies of an image, the use of a fixed-length code created by the overlap conversion makes it possible to obtain a code other than the code to be decoded. In addition, the code required for reference must be accessed, so that the transfer efficiency inside the device is reduced to a fraction. If a large number of buffer memories are provided to avoid this, the number of accesses is reduced and the transfer efficiency is improved, but the memory cost is increased this time. Under such circumstances, it is desirable to use non-overlapping transforms. Non-overlapping transform means include a discrete cosine (DCT) transform and a Haar wavelet transform. The Haar wavelet transform is particularly preferable because of its ease of transformation.

The AC component quantization section 508 subjects the input AC components HR, HG, and HB of each component to first-stage quantization and outputs the result to the color conversion section 509. In the first-stage quantization in the AC component quantization unit 508, for example, HH is deleted from the AC components HL, LH, and HH. This is because the HH coefficient is a coefficient corresponding to the oblique edge of the image and has the lowest visual significance among the Haar coefficients. Therefore, even if this coefficient is reduced, the image quality deterioration is not so noticeable.

The color conversion unit 509 includes an AC component quantization unit 50
8 is subjected to color conversion (brightness / color separation) of the AC component of each component that has been subjected to the first stage quantization and
An AC lightness signal YH and AC color signals UH and VH are generated and output to the quantization unit 510. Further, the color conversion unit 507
Color, G, and B color position correlation conversion units 504, 505, and 506
DC components LLR, L of each component input from
LG and LLB are color-converted (lightness / color separation) to generate DC lightness signal YLL and DC color signals ULL and VLL, and output them to quantization section 510.

The color conversion performed by the above-described color conversion units 507 and 509 is a YUV conversion using the equation shown in FIG. In FIG. 5, among the obtained coefficients, Yz is an “AC lightness signal”, and Uz and Vz are “AC color signals”. Note that the color conversion of the present invention is not limited to this, and Lab, YI
A similar effect can be obtained by using a conversion of Q, YCrCb, or the like. These conversions are usually performed on the RGB color signals, and as a result, the color signals become 0 in the achromatic image.
Also when applied to the AC component of each color of GB, the value corresponding to the color signal is 0 in an achromatic image. Under such circumstances, in this specification, not only the case of performing the conversion on the RGB signal but also the general conversion of the image components such as the AC component and the DC component is referred to as “the conversion into the brightness signal and the color signal”. ".

The quantizing section 510 includes a DC lightness signal YLL input from the color converting section 507 and DC color signals ULL and VL.
L is quantized, a second-stage quantization is performed on the AC lightness signal YH and the AC color signals UH and VH input from the color conversion unit 509, and the result is output to the encoding unit 511.

As for the quantization of the “DC lightness signal” YLL and the “DC color signal” ULL and VLL, the “DC lightness signal” YLL is linearly quantized to 5 bits by truncating bits, and the “DC color signal” ULL, VL
L is quantized to 5 bits by non-linear quantization. Here, the non-linear quantization is "DC color signal"
It is assumed that a portion having a small absolute value of ULL and VLL is finely quantized, and a portion having a large absolute value is coarsely quantized. The reason for performing such non-linear quantization is as follows: “DC color signal” ULL,
In VLL, information is concentrated around 0 by color conversion,
In addition, a slight change in the numerical value near 0 greatly affects the tint. Therefore, unless the value near 0 is finely quantized, visual deterioration becomes conspicuous. FIG.
FIG. 38 to FIG. 38 are diagrams illustrating an example of a non-linear quantization table in the case of non-linearly quantizing the “DC color signals” ULL and VLL.

Yz (z = H) which is an “AC lightness signal”
L, LH, HH) are vector-quantized to obtain an "AC color signal".
Uz and Vz are vector quantized and then Y
Assume that the ratio of the z to the vector is encoded.

First, Yz (z = z = AC lightness signal)
HL, LH, HH) will be described in detail. First, a vector YH = “AC brightness signal”
After creating (YHL, YLH) (in this embodiment, YHH is always 0 because HH is reduced in the AC component quantization unit 508), the “AC brightness signal” is obtained according to the quantization table of FIG. Vector YH = (Y
HL, YLH). The quantization table shown in FIG. 6 shows a case where vector quantization is performed on 13-valued 4 bits.

Here, the advantage of performing vector quantization is that the coding efficiency is much better than when scalar quantization is performed for each AC component. For example, when it is desired to leave the vertical and horizontal edges important for image quality without deteriorating, in the case of vector quantization, as shown in FIG. ,
In the case of scalar quantization, as shown in FIG.
Vectors requiring as much as 9 bits and 6 bits of information and having very low appearance frequencies in 49 values (for example, (YHL, YL)
H) = (64, 64), etc., both YHL and YLH have large values), so that the coding efficiency deteriorates. In the vector quantization, specifically, the distance between all the vectors and the vector YH is calculated, and the vector having the smallest distance is selected.

When the high-frequency component is vector-quantized, if a vector whose length is shorter than that before quantization is selected, moiré (periodic pattern deterioration) of a halftone dot portion hardly occurs. . For example, for a vector (44, 0, 0), quantization value candidates are (48, 0, 0) and (3
When (2, 0, 0) is (48,
Even when the quantization value (32, 0, 0) is selected, moire does not occur in the halftone dot portion. Therefore, when considering the moiré of the halftone dot portion, the quantization threshold may be set to be larger than the intermediate point of the quantization value.

Next, Uz and Vz which are "AC color signals"
Will be described. When quantizing the vectors of Uz and Vz, which are “AC color signals”, the ratio with the vector of Yz is encoded as described above. This is a quantization method using the fact that the structures of the “AC lightness signal” and the “AC color signal” are similar in the same block.

Vector quantization of Uz and Vz, which are “AC color signals”, will be described with reference to FIG. FIG. 8 is an explanatory diagram for explaining vector quantization of Uz and Vz that are “AC color signals”.

For example, when there are two different color boundaries in the vertical direction in the block as shown in (1100) in FIG. 8, the information (1101) of each color of RGB is Haar transformed (1102), Color conversion of AC component (1103)
And the “AC lightness signal” vector and the “AC color signal” vector are in a proportional relationship with each other. That is, YH = nu × U
H, YH = nv × UH. However, YH, UH, VH
Is a vector, YH = (YHL, YLH), UH
= (UHL, ULH), VH = (VHL, VLH), and nu and nv are constants. Note that the two vectors are two-dimensional vectors because the HH component is 0 in this example, but originally, the three-dimensional vector YH = (YH
L, YLH, YHH).

Therefore, if the proportional coefficients nu and nv are quantized (1105), efficient quantization can be performed. Such a proportional relationship between vectors is always obtained when the number of different colors in a block is two or less, and 3 in a small block of 2 × 2 pixel blocks.
This is an effective quantization method, since it is rare that a mixture of colors equal to or more than one color is used. However, in practice, slight fluctuations occur in colors due to scan fluctuations and the like, and in many cases, a perfect proportional relationship does not hold. Therefore, in the present embodiment, the ratio is encoded as follows.

That is, first, YH is vector-quantized, and the angle θ between YH and UH is obtained by inner product calculation.
If θ is closer to 0 degrees or 180 degrees than the threshold value, it is considered that YH and UH are proportional. If θ is closer to 0 degrees, | YH | / | UH | In the case where the value is close to, − | YH | / | UH | is set to the proportional value nu. Here, for example, | YH | is a vector YH =
(YHL, YLH) means | YH | = √
(YHL * YHL + YLH * YLH). Further, an appropriate threshold value differs depending on the machine accuracy, and it is difficult for a machine having a large scan fluctuation or the like to be regarded as having a proportional relationship, so that the threshold value needs to be set to a relatively large value. θ is YH and U
If H is a value that cannot be regarded as proportional, it is encoded with the proportionality coefficient set to 0. Finally, the proportional coefficient is quantized according to the table of FIG. The same applies to V. As shown in FIG. 9, the proportional coefficient is stored in 5 bits of 25 values each including U and V.

The encoding section 511 creates a fixed-length code by combining the information (in units of 2 × 2 pixels) quantized by the quantization section 510 and stores it in the page memory 403. FIG. 10 shows an example of a fixed-length code format. As shown in FIG. 10, as a result of quantization by the quantization / encoding units 510 and 511, the “DC lightness signal” YLL and the “DC color signal” UL
L and VLL are grouped into 5 bits, (vector) YH is grouped into 4 bits, (vector) UH and (vector) VH are grouped into 5 bits, and when they are combined, a fixed length code of 24 bits can be obtained. Since the image before compression was 4 pixels × 3 colors × 8 bits = 96 bits, the compression rate of ４ was achieved in this embodiment.

As described above, according to the encoding unit of the first embodiment, the position correlation conversion is performed for each of the R, G, and B components of the image data (for example, 8 bits per pixel for each color). Since the color conversion is performed by the color conversion unit, the coding circuit can be simplified and the quantization can be simplified.

According to the encoding unit of the first embodiment,
Since we decided to perform Haar transform as position correlation transformation,
Position / phase conversion can be performed with a simple configuration. Further, the AC component quantization unit 508 deletes the coefficient (HH coefficient) corresponding to the oblique edge of the image.
Since this coefficient is the least visually significant among the Haar transforms, it has the effect that even if it is deleted, it has little effect on image quality degradation.

Hereinafter, the effect of the encoder 402 according to the first embodiment will be specifically described in comparison with the related art. The encoder 402 according to the first embodiment has an advantage that “a strong bias of information occurs upstream of processing” as compared with the related art. That is, since the positional correlation of the image is generally stronger than the correlation between the colors of the components, if the conversion using the positional correlation is performed upstream of the processing, various quantizations can be performed at that time.

More specifically, the AC component quantization unit 50 shown in FIG.
In H.8, the HH coefficients are quantized first. The first-stage quantization can reduce the number of color conversion circuits for the HH coefficients. FIG. 11A shows the required numbers of the conventional color conversion circuits and Haar conversion circuits. The color conversion circuit requires four circuits for 2 × 2 pixels, and the Haar conversion circuit requires three circuits for YUV. On the other hand, FIG. 11B shows the case of this embodiment. Since the HH coefficient is reduced by quantization, it is not necessary to perform color conversion for the HH coefficient, and the number of color conversion circuits can be reduced by one.

Further, as described above, the quantization of the AC component quantization unit 508 is not performed by deleting one coefficient as a whole.
Even if only the lower bits of the coefficient are reduced, the number of bits handled by the color conversion circuit can be reduced, and the processing efficiency can be improved. For example, the lower bits of the HL, LH, and HH coefficients originally having a depth of 9 bits, 9 bits, and 10 bits are represented by H
If the color conversion is performed after reducing the L and LH coefficients by 2 bits and the HH coefficient by 3 bits, the number of bits handled by the color conversion circuit is only 7 bits, 7 bits, and 7 bits.
FIG. 11C shows the required numbers of color conversion circuits and Haar conversion circuits in this case.

On the other hand, the conventional method, for example,
According to the method disclosed in Japanese Patent No. 245070, color conversion is first performed, and then a brightness signal and a color signal are decomposed into a DC component and an AC component. It has an important meaning, and if the lower bits are deleted, a problem such as disappearing of the color of a light background occurs, so that quantization cannot be performed before color conversion. For example, when the color signals U and V each have a value of “3”, if the Haar transform is performed after performing 2-bit quantization, the color will disappear at the time of quantization. In order to avoid this problem, it is conceivable to use non-linear quantization for the U and V coefficients in the conventional method. In this case, however, the Haar wavelet transform coefficient obtained in the subsequent stage becomes sensitive to the quantization, However, there is a problem that efficiency is poor because the conversion leads to large deterioration.

Generally, in order to prevent color deterioration, the DC component of a color signal may be stored. In the present embodiment, the DC component is color-converted after being quantized to some extent while the DC component is color-converted accurately while maintaining the tint while the AC component is maintained.
Circuit size can be reduced.

As described above, in this embodiment, the effect of reducing the circuit scale can be obtained without lowering the coding efficiency as compared with the conventional method. Here, “without lowering the coding efficiency as compared with the conventional method” means that an information localization effect equivalent to the conventional method can be obtained in that both the positional correlation and the color correlation are used. In particular, in the case of the present embodiment, since both the Haar transform and the color transform are linear transforms, the values of the transform coefficients are exactly the same irrespective of the order of the two transforms without considering the quantization.

FIG. 12 shows Embodiment 1 of the decoder 404.
FIG. A decoder 404 according to the first embodiment shown in FIG. 12 is for decoding the compressed code encoded by the encoder 402 according to the first embodiment. Figure 1
2 includes a code division unit 521, an inverse quantization unit 522, an inverse color conversion unit 523, and an inverse color conversion unit 52.
4. AC component inverse quantization unit 525, R color position inverse correlation unit 526, G color position inverse correlation unit 527, B color position inverse correlation unit 528, restored R color nxm buffer 529, restored G A color nxm buffer 530 and a restored B color nxm buffer 531 are provided.

Next, the decoding method of the decoder 404 in FIG. 12 will be described in detail. A compressed code compressed by the encoder 402 and stored in the page memory 403 (see FIG. 10)
Is read from the page memory 403 in units of 2 × 2 pixel blocks at the time of image formation, and is decoded by the decoder 404.
After being subjected to filter processing and the like by the filter / gradation processing unit 405, the data is transferred to the writing system.

The code division unit 521 is provided in the page memory 403
The compression code of the 2 × 2 pixel block stored in
YLL, ULL, VL according to the code configuration.
The signal is distributed to L, YH, UH, and VH, and output to the inverse quantization unit 522. The inverse quantization unit 522 includes a compression code YLL, ULL, VLL, YH, U
H and VH are inversely quantized.

More specifically, the inverse quantization unit 522 calculates
By adding 3 bits to the lower part of 5 bits to make 8 bits, U
LL and VLL are restored to 9 bits by using a table for performing inverse transform of nonlinear quantization on the encoding side. In addition,
Here, when adding 3 bits to the lower order of 5 bits of YLL, simply adding “000” by bit shift lowers the density, so it is desirable to add an intermediate value of “100”. To explain this point in detail, for example, 23 = 1
Delete lower 2 bits of value 0111 (divide by 4)
And the lower 2 bits are 00 when decoding.
Is added, 10100 = 20. However, adding the intermediate value 2 of 4 to the lower order results in 10110 = 23. In general, restoring the intermediate value results in less change in density and improvement in the S / N ratio. It becomes possible.

The inverse quantization unit 522 restores the vector using the table shown in FIG. 6 according to the code value for YH. For UH and VH, nu and nv are restored from the 25-value code in the reverse procedure to that at the time of encoding.

The inverse color conversion unit 523 performs inverse conversion of YUV conversion on the restored DC lightness signal YLL, DC color signals ULL, VLL, and converts the restored LLR, LLG, LLB into an R color position correlation inverse conversion unit 526. , G color position correlation inverse transform unit 52
7. Output to the B color position correlation inverse transform unit 528, respectively.

The inverse color conversion section 524 performs the inverse conversion of the YUV conversion on the restored AC lightness signal YH and AC color signals UH and VH, and outputs the restored HR, HG and HB to the AC component inverse quantization section 525. Output. The AC component inverse quantization unit 525 inversely quantizes the restored HR, HG, and HB (in this case, adds HH = 0 to each color), and converts the R color position correlation inverse transform unit 52.
6. Output to the G color position correlation inverse conversion unit 527 and the B color position correlation inverse conversion unit 528, respectively.

R, G, B color position correlation inverse conversion section 52
6, 527, 528 are restored (LLR, HR),
(LLG, HG) and (LLB, HB) are each subjected to the inverse transform of the Haar wavelet transform to obtain a restored R color and a restored G
The color and the restored B color are stored in the nxm buffers 529, 530 and 531, respectively. As described above, restoration is performed in units of 2 × 2 pixel blocks. If a large amount of memory can be used, the processing time can be reduced by performing each inverse conversion in a table reference method.

As described above, according to the decoding unit of the first embodiment, it is possible to restore the code information encoded by the encoder 402 of the first embodiment.

(Embodiment 2) FIG.
FIG. 9 is a block diagram showing a second embodiment of the present invention. The encoder 402 according to the second embodiment is different from the encoder 402 according to the first embodiment.
The difference from FIG. 2 is that the encoder 402 of the first embodiment is different.
Although the configuration is such that the DC component is separated into the brightness signal and the color signal, the encoder 402 according to the second embodiment does not separate the DC component into the brightness signal and the color signal. The encoder 402 according to the second embodiment has a configuration in which the color conversion unit 507 is deleted from the first embodiment (FIG. 2), and the functions of other parts are almost the same as those of the first embodiment (FIG. 2). The same is true.

The encoder 402 shown in FIG. 13 has R, G, and B components (for example, 8 bits per pixel for each color) input from the scanner 401 in n × m units (for example, 2 × 2 pixel block units). ), N, m buffers 601 of R, G, and B colors for temporarily storing image data
602 and 603 and n × m buffers 60 for R, G and B colors
1, 602, and 603 (R
GB), a conversion using position correlation (hereinafter referred to as “position correlation conversion”) is performed to separate a DC component and an AC component for each color, and the DC component is output to the quantization unit 610, while the AC component is converted to an AC component. R color, G color, B output to the component quantization unit 608
Color position correlation conversion units 604, 605, and 606 are provided.

The encoder 402 shown in FIG.
Color, G, B color position correlation conversion units 604, 605, 606
The AC component quantization unit 608 that performs the first-stage quantization on the AC component of each component input from the first component and outputs the result to the color conversion unit 609, and the first-stage quantization that is input from the AC component quantization unit 608. A color conversion unit 609 that performs color conversion (brightness / color separation) on the quantized AC component of each component to generate an AC lightness signal and an AC color signal, and outputs the signal to the quantization unit 610; G color, B color position correlation conversion unit 60
4, 605, and 606, and quantizes the AC signal and the AC color signal input from the color conversion unit 609 in the second stage. And a quantization unit 610 that outputs the
An encoding unit 611 that creates a fixed-length code by combining the information (in units of 2 × 2 pixels) quantized in 10 and stores the code in the page memory 403.

Next, the encoding method of the encoder 402 shown in FIG. 13 will be described in detail. Image data composed of R, G, and B color components (for example, 2 × 2 pixel block units, each pixel of 8 bits) read from the scanner 401 is composed of R × G, B color n × m buffers 601 and 60.
2, 603 and stored temporarily.

The R, G, and B color position correlation conversion unit 604,
Reference numerals 605 and 606 denote n × m buffers 60 for R, G, and B colors.
1, 602, and 603 (R
GB) to perform a position correlation conversion to obtain a DC component L for each color.
LR, LLG, LLB and AC components HR, HG, HB are separated, and the DC components LLR, LLG, LLB are quantized by the quantization unit 61.
While outputting to 0, the AC components HR, HG, HB are output to the AC component quantization unit 608. As the position correlation conversion,
As in the first embodiment, the Haar transform is used.

The AC component quantization section 608 subjects the input AC components HR, HG, HB of each component to quantization at the first stage, and outputs the result to the color conversion section 609. In the first-stage quantization in the AC component quantization unit 608, H having a depth of 9 bits, 9 bits, and 10 bits is originally used.
The lower bits of the L, LH, and HH coefficients are reduced by 2 bits for the HL and LH coefficients and 3 bits for the HH coefficient (see FIG. 11C).

The color conversion unit 609 includes an AC component quantization unit 60
8 is subjected to color conversion (brightness / color separation) of the AC component of each component that has been subjected to the first stage quantization and
An AC lightness signal YH and AC color signals UH and VH are generated and output to the quantization unit 610. For color conversion, YUV conversion is performed as in the first embodiment.

The quantization section 610 quantizes the DC signals LLR, LLG, LLB inputted from the R, G, B color position correlation conversion sections 604, 605, 606, and
11 is output. Specifically, the quantization unit 610 outputs
The DC components LLR, LLG, and LLB of B are linearly quantized into 5 bits each. Further, the quantization unit 610 includes a color conversion unit 609.
AC brightness signal YH and AC color signals UH, V
H is subjected to a second-stage quantization, and the encoding unit 6
11 is output. The second stage quantization performs the same vector quantization as in the first embodiment.

The coding section 611 creates a fixed-length code by combining the information (in units of 2 × 2 pixel blocks) quantized by the quantization section 610 and stores it in the page memory 403. Figure 1
5 shows an example of a fixed-length code format. As in the first embodiment, a 24-bit fixed length code is created.

As described above, according to the encoding unit of the second embodiment, since the color conversion is not performed on the DC component of each component, the effect of the first embodiment (“the strong information Bias occurs upstream of processing "), and the following circuit scale reduction effect is achieved.

A quantization table for quantizing the DC component of the color signal becomes unnecessary. The smaller the absolute value of the DC component of the color signal is, the more significant the value is. Therefore, it is usually necessary to perform nonlinear quantization when quantizing the DC component. FIG. 14 shows an example of a non-quantized quantization table for performing such non-linear quantization. As a method of performing nonlinear quantization, there are a method of performing calculation and a method of using a conversion table. However, the former has a slower processing speed, and the latter has a faster processing speed, but requires a conversion table in the machine, which is costly. Such difficulty is caused by quantizing DC components having two different properties, that is, a brightness signal and a color signal. On the other hand, in the encoding unit according to the second embodiment, since the DC component of each color is not subjected to color conversion, quantization of the DC component can be all performed by a simple bit reduction method. There is no need to prepare a table or the like.

Since the color conversion of the DC component is not performed, the color conversion circuit for the DC component becomes unnecessary. FIG. 16 illustrates the required numbers of color conversion circuits and Haar conversion circuits in the encoding unit according to the second embodiment. As shown in the figure, a color conversion circuit for a DC component is not required.

In the encoding unit according to the second embodiment, since the DC component is not subjected to color conversion, the localization effect of the finally obtained information is slightly weak for the DC component. But,
In the case of fixed-length compression, this disadvantage is not necessarily large. Because, in a color copier that requires high-definition color reproduction, it is often not possible to significantly remove the DC component of color signals, even in terms of image quality. This is because the number of bits of the component does not change significantly.

In the commonly used encoding method, the AC component of the color signal may be discarded (so-called subsampling), but the DC component is not quantized much. For this reason, in many cases, the merit of circuit scale reduction such as the reduction of one nonlinear quantization table and one color conversion circuit is larger than the disadvantage of not performing color conversion of a DC component.

In the encoder 402 according to the second embodiment, the AC component quantization section 608 may be omitted. In this case, although there is no effect due to the “strong bias of information upstream” described in the description of the first embodiment, the above-described effect can be obtained.

FIG. 17 shows a second embodiment of the decoder 404.
FIG. A decoder 404 according to the second embodiment shown in FIG. 17 is for decoding the compressed code encoded by the encoder 402 according to the second embodiment. Figure 1
7 includes a code division unit 621, an inverse quantization unit 622, an inverse color conversion unit 624, an AC component inverse quantization unit 625, an R color position correlation inverse conversion unit 626, and a G color position correlation An inverse transform unit 627, a B color position correlation inverse transform unit 628, a restored R color nxm buffer 629, a restored G color nxm buffer 630, and a restored B color nxm buffer 631 are provided.

Next, the decoding method of the decoder 404 in FIG. 17 will be described. The compressed code (see FIG. 15) compressed by the encoder 402 and stored in the page memory 403 is read from the page memory 403 in units of 2 × 2 pixel blocks at the time of image formation, and decoded by the decoder 404. Then, after being subjected to filter processing and the like by the filter / gradation processing unit 405, the data is transferred to the writing system.

The code division unit 621 converts the compression code of the 2 × 2 pixel block (see FIG. 15) according to the code configuration.
LLR, LLG, LLB, (vector) YH, (vector) UH (vector) VH, and the inverse quantization unit 62
Output to 2. The inverse quantization unit 622 includes a code division unit 621
The compression codes LLR, LLG, LLB, (vector) YH and (vector) UH (vector) VH input from are dequantized.

More specifically, the inverse quantization unit 622 calculates LL
The R, LLG, and LLB are restored to 9 bits and output to the R color position correlation inverse transform unit 626, G color position correlation inverse transform unit 627, and B color position correlation inverse transform unit 628, respectively. In addition, the inverse quantization unit 622 determines the YH according to the code value,
And reconstruct the vector using the table in UH,
With regard to VH, nu and nv are restored from the 25-value code in the reverse procedure to that at the time of encoding, and output to the inverse color conversion unit 624.

The inverse color conversion unit 624 performs the inverse conversion of the YUV conversion on the restored AC lightness signal YH and AC color signals UH and VH, and outputs the restored HR, HG and HB to the AC component inverse quantization unit 625. Output. The AC component inverse quantization unit 625 inversely quantizes the restored HR, HG, HB (in this case, adds HH = 0 to each color), and converts the R color position correlation inverse transform unit 62.
6. Output to the G color position correlation inverse conversion unit 627 and the B color position correlation inverse conversion unit 628, respectively.

R, G, B color position correlation inverse conversion section 62
6, 627, 628 are restored (LLR, HR),
(LLG, HG) and (LLB, HB) are each subjected to the inverse transform of the Haar wavelet transform to obtain a restored R color and a restored G
The color and the restored B color are stored in the nxm buffers 629, 630 and 631. As described above, restoration is performed in units of 2 × 2 pixel blocks. When a large amount of memory can be used, the processing time can be reduced by performing each inverse conversion by a table reference method.

As described above, according to the decoding unit of the second embodiment, it is possible to restore the code information coded by the encoder 402 of the second embodiment.

(Embodiment 3) FIG.
FIG. 13 is a block diagram showing a third embodiment of the present invention. The encoder 402 according to the third embodiment is different from the encoder 402 according to the first embodiment.
The difference from FIG. 2 is that the encoder 402 of the first embodiment is different.
Although the configuration is such that the AC component is separated into the brightness signal and the color signal, the encoder 402 according to the third embodiment quantizes the AC component without separating the AC component into the brightness signal and the color signal. And The encoder 402 according to the third embodiment has a configuration in which the color conversion unit 509 is deleted from the encoding unit (FIG. 2) according to the first embodiment, and the functions of the other units are the same as those according to the first embodiment.
Is substantially the same as the encoder 402 (FIG. 2).

The encoder 402 shown in FIG. 18 has R, G, and B components (for example, 8 bits per pixel for each color) input from the scanner 401 in n × m units (for example, 2 × 2 pixel block units). ), An n × m buffer 701 of R, G, and B colors for temporarily storing image data
702, 703 and n × m buffers 70 of R, G, B colors
1, (702, 703)
GB) for each color, to separate a DC component and an AC component for each color, and to output the DC component to the color conversion unit 707, while outputting the AC component to the AC component quantization unit 708; G color, B color position correlation conversion units 704, 705, 70
6 is provided.

Further, the encoder 402 shown in FIG.
Color, G, and B color position correlation conversion units 704, 705, and 706
The DC component of each component input from the color conversion unit is color-converted, separated into a DC lightness signal and a DC color signal, and
The AC component of each component input from the color conversion unit 707 outputting to 0 and the R, G, and B color position correlation conversion units 704, 705, and 706 is subjected to a first-stage quantization to perform a quantization process. An AC component quantization unit 708 that outputs the signal to an AC component quantizing unit 710;
The DC brightness signal and the DC color signal input from the color conversion unit 707 are quantized, and the AC signal input from the AC component quantization unit 708 is subjected to a second-stage quantization,
A quantizing unit 710 that outputs to the encoding unit 711 and an encoding unit 711 that creates a fixed-length code by combining the information (in units of 2 × 2 pixels) quantized by the quantization unit 710 and stores the code in the page memory 403. And

Next, the operation of the encoder 402 shown in FIG. 18 will be described in detail. In FIG. 18, image data composed of components of R, G, and B colors (for example, 2 × 2 pixel block units, 8 bits for each pixel of each color) read from the scanner 401 is nxm of R, G, and B colors. Buffer 7
01, 702, and 703, and are temporarily stored.

The R, G, and B color position correlation conversion unit 704,
Reference numerals 705 and 706 denote n × m buffers 70 for R, G, and B colors.
1, (702, 703)
GB) to perform a position correlation conversion to obtain a DC component L for each color.
LR, LLG, LLB and AC components HR, HG, HB are separated, and the DC components LLR, LLG, LLB are separated into color conversion units 70.
7, while the AC components HR, HG, HB are output to an AC component quantization unit 708. As the position correlation conversion,
As in the first embodiment, the Haar transform is used.

The AC component quantization section 708 subjects the input AC components HR, HG, HB of each component to first-stage quantization and outputs the result to the quantization section 710. In the first-stage quantization in the AC component quantization unit 708, an H having a depth of 9 bits, 9 bits, and 10 bits is originally used.
The lower bits of the L, LH, and HH coefficients are reduced by 2 bits for the HL and LH coefficients and 3 bits for the HH coefficient (see FIG. 11C).

The color conversion unit 707 performs color conversion (brightness / color separation) on the DC components LLR, LLG, and LLB of each component input from the R, G, and B color position correlation conversion units 704, 705, and 706. Then, a DC lightness signal YLL and DC color signals ULL and VLL are generated and output to the quantization unit 710.

The quantizing section 710 includes a DC lightness signal YLL input from the color converting section 707 and DC color signals ULL and VL.
L is quantized, and the AC signals HR, HG, and HB input from the AC component quantization unit 708 are subjected to the second-stage quantization, and are output to the encoding unit 711. here,
As the second stage quantization of the AC component, the G color AC component HG is vector-quantized, and the R and B color AC components H
For R and HB, the ratio with the AC component of G color is quantized.

The coding section 711 creates a fixed-length code by combining the information (in units of 2 × 2 pixels) quantized by the quantization section 710 and stores it in the page memory 403. FIG. 19 shows an example of a fixed-length code format. As in the first embodiment, a 24-bit fixed length code is created.

As described above, according to the encoding unit of the third embodiment, since the AC component is not separated into the brightness signal and the chrominance signal, the effect of the first embodiment (strong bias of information at the upstream). In addition, it is possible to reduce the circuit scale that eliminates the need for a color conversion circuit for AC components. FIG.
19 illustrates the required numbers of color conversion circuits and Haar conversion circuits in the encoding unit according to the third embodiment. As shown in the figure,
A color conversion circuit for an AC component is not required.

Here, the proportional relationship between the AC component vectors existing between the “AC lightness signal” and the “AC color signal” in the first embodiment is the proportional relationship between the AC component vectors of the R, G, and B colors. It is derived from. Therefore, this proportional relationship always holds when there are only two or less colors in the block as described in the first embodiment. Therefore, the first embodiment can be used even if the proportional relationship between AC components is used.
In the same manner as described above, the AC component can be efficiently compressed.

In the third embodiment, the G color AC component HG is vector-quantized by the quantization section 710, and the ratio of the R and B color AC components HR and HB to the G color AC component is determined. Although quantization is performed, if the G color has no edge, but the R and B colors have an edge, there is a problem that the edges disappear in all three colors. In order to solve this, for example, the color having the largest edge (referred to as a specific color) among the three colors may be vector-quantized, and the ratio of the other colors to the vector of the specific color may be quantized.

In this case, 2 bits are added to the fixed length code, and
It is necessary to indicate which color is a specific color. Also, instead of 2 × 2 block units, larger blocks,
For example, a specific color can be determined in units of 4 × 4 blocks. In this case, since a code indicating a specific color of 2 bits is attached to four 2 × 2 blocks, four 2 × 2 blocks are consequently obtained.
A fixed-length code is created in units of two blocks.

(Embodiment 4) FIG. 21 is a block diagram showing Embodiment 4 of the encoding unit in FIG. In the fourth embodiment, the encoding unit (FIG. 2) of the first embodiment uses R color,
A case will be described where the G and B color position correlation conversion units 504, 505, and 506 perform frequency conversion (Haar conversion and the like) as position correlation conversion, but perform BTC conversion.

FIG. 21 is a diagram showing Embodiment 1 of the encoder 402.
FIG. Encoder 402 shown in FIG.
Are n × m buffers 801, 802, 8 of R, G, B colors
03 and R, G, and B color position correlation conversion units 804 and 80
5, 806, a color conversion unit 807, a color conversion unit 809, a quantization unit 810, and an encoding unit 811.

The BTC conversion is described in JP-A-10-257331.
As shown in paragraph number (0151) of the publication,
This is a known fixed-length compression technology. For example, an image is quantized for each block of 4 × 4 pixels by average density data LA (8 bits), gradation width index data LD (8 bits), and each pixel value in 4 × 4 blocks. This is a method of converting into quantized value data φ (2 bits each) when it is converted.

Next, an encoding method of the encoder 402 in FIG. 21 will be described. R color, G color, B color n in 4 × 4 pixel units
The pixel data of each color read into the × m buffers 801, 802, and 803 are respectively BTC-converted by the R, G, and B color position correlation conversion units 804, 805, and 806, and the L
A, LD, and φ are generated. Hereinafter, the LA coefficient of the X color is represented by L
The LD coefficients of AX and X are abbreviated as LDX, and the φ coefficient of X is abbreviated as φX (X = R, G, B).

The color conversion unit 807 is input from the R, G, and B color position correlation conversion units 804, 805, and 806 (L
AR), (LAG, LDG), and (LAB) by color-converting the LD into (YLA, YLD), (ULA, VLA),
(ULD, VLD) are output to the quantization unit 810. Also, the color conversion unit 809 is input from the R, G, and B color position correlation conversion units 804, 805, and 806 (φR),
(ΦG) and (φB) are color-converted into (Yφ), (Uφ,
Vφ) to the quantization unit 810.

The color conversion is performed by the following equation, and the coefficients after the color conversion are abbreviated as follows. YZ = (ZR + 2 * ZG + ZB) / 4 UZ = ZR-ZG VZ = ZB-ZG (Z = LA, LD, φ)

The quantizer 810 receives the input (YL
A, YLD), (ULA, VLA), (ULD, VL
D), (Yφ), and (Uφ, Vφ) are quantized and output to the encoding unit 811. Specifically, the encoding unit 811 for YLA and YLD is not quantized with each bit being 8 bits.
Sent to. The ULA, ULD, VLA, and VLD are quantized to 6 bits in the same manner as the non-linear quantization of ULL and VLL to 6 bits in a non-edge portion shown in FIG. In addition, Yφ is sent to the encoding unit 811 without being quantized, leaving 32 bits. Uφ and Vφ are 1
If it is treated as a 6-dimensional vector and the absolute value of the normalized inner product with Yφ is sufficiently close to 1, for example, for Uφ, | Yφ · Uφ |> 0.8 × | Yφ || Uφ | Assuming that Yφ = nu × Uφ, the proportional constant nu = | Uφ | / | Yφ | is quantized. On the other hand, when | Yφ · Uφ | is 0.8 or less,
Assuming that it is not proportional, quantization is performed with nu = 0.
The same applies to Vφ. The proportionality constants nu and nv are respectively -2.0, -1.98,-
1.96, ..., 0, ..., 1.96, 1.98,2.0
And an 8-bit value of 201 values divided into 0.02 units.

The encoding section 811 creates a fixed-length code by combining the information (in units of 4 × 4 pixel blocks) quantized by the quantization section 810 and stores it in the page memory 403. Thus, YLA, YLD = 8 bits each, ULA, ULD,
VLA, VLD = 6 bits each, Yφ = 32 bits, U
Since each of φ and Vφ is coded as 8 bits, the 4 × 4 pixel block is compressed to 88 bits in the entire fixed length code, and a fixed length compression of about の is achieved.

According to the encoding unit of the fourth embodiment, BTC
Although the quantization itself is not provided before the color conversion because the conversion itself includes the quantization, the BT which is the position correlation conversion on the upstream side of the processing is used.
Since C is performed, the number of bits of φ sent to the color conversion unit is reduced to 2 bits. Thereby, the number of input bits of the color conversion circuit can be reduced.

FIG. 22 shows Embodiment 4 of the decoder 404.
FIG. A decoder 404 according to the fourth embodiment shown in FIG. 22 is for decoding the compressed code encoded by the encoder 402 according to the fourth embodiment. FIG.
2 includes a code division unit 821, an inverse quantization unit 822, an inverse color conversion unit 823, and an inverse color conversion unit 82.
4. R color position correlation inverse conversion section 826, G color position correlation inverse conversion section 827, B color position correlation inverse conversion section 828,
G and B color nxm buffers 829, 830, 831 are provided.

The decoding method of the decoder 404 shown in FIG. 22 will be described. The code division unit 821 is used for the page memory 403
The code information (4 × 4 pixel block unit) stored in
(YLA, YLD), (ULA, VLA), (ULD,
VLD), (Yφ), and (Uφ, Vφ) and output to the inverse quantization unit 822. The inverse quantization unit 822 calculates (YL
A, YLD), (ULA, VLA), (ULD, VL
D), (Yφ), (Uφ, Vφ) are inversely quantized, and (YLA, YLD), (ULA, VLA),
(ULD, VLD) is output to the inverse color conversion unit 823, and (Yφ) and (Uφ, Vφ) after inverse quantization are output to the inverse color conversion unit 824.

The inverse color conversion unit 823 receives the input (YL)
A, YLD), (ULA, VLA), (ULD, VL
D), inverse color conversion of the color conversion of the color conversion unit 807 is performed to restore (LAR), (LAG, LDG), and (LAB), and the R color position correlation inverse conversion unit 826, G color The signals are output to the position correlation inverse transform unit 827 and the B color position correlation inverse transform unit 828 respectively. Also, the inverse color conversion unit 824 receives the input (Y
φ) and (Uφ, Vφ) are subjected to inverse color conversion by the inverse color conversion unit 804 to restore (φR), (φG), and (φB), and the R color position correlation inverse conversion unit 826, The signals are output to the G color position correlation inverse transform unit 827 and the B color position correlation inverse transform unit 828, respectively.

The R color position correlation inverse conversion section 826, G color position correlation inverse conversion section 827, and B color position correlation inverse conversion section 828
(LAR) and (φR), (LAG, LDG) and (φ
G), (LAB) and (φB) are subjected to inverse BTC conversion, and stored in the restored R color, restored G color, restored B color n × m buffers 829, 830, 831. As described above, restoration is performed in units of 4 × 4 pixel blocks.

According to the decoding unit of the fourth embodiment, it is possible to restore the code information encoded by the encoder 402 of the fourth embodiment.

(Embodiment 5) FIG.
FIG. 16 is a block diagram showing a second embodiment 5; The encoder 402 according to the fifth embodiment is different from the encoder 402 according to the first embodiment.
(FIG. 2) is obtained by adding an area determination unit.

An encoder 402 shown in FIG. 23 has R, G, and B components (for example, 8 bits per pixel for each color) input from the scanner 401 in n × m units (for example, 2 × 2 pixel block units). ), Nxm buffers 901 of R, G, and B colors for temporarily storing image data
902, 903 and R × G, B color n × m buffer 90
1, 902, and 903 (R
GB) for each color, to separate a DC component and an AC component for each color, output the DC component to the color conversion unit 907, and output the R color to output the AC component to the AC component quantization unit 908; G color, B color position correlation conversion units 904, 905, 90
6 and the R, G, and B color position correlation conversion units 904 and 90
5, a color conversion unit 9 that performs color conversion (brightness / color separation) of the DC component of each component input from the components 906 to generate a DC lightness signal and a DC color signal, and outputs the signal to the quantization unit 910.
09.

The encoder 402 shown in FIG.
Color, G, B color position correlation conversion units 904, 905, 906
, An AC component quantizer 908 that performs a first-stage quantization of the AC components of the components input from the first component and outputs the result to the color converter 909, and a first-stage quantizer that is input from the AC component quantizer 908. The AC component of each of the quantized components is subjected to color conversion (lightness / color separation) to generate an AC lightness signal and an AC color signal, and output the AC lightness signal and the AC color signal to the quantization unit 910. At the same time, a color conversion unit 909 that outputs an AC brightness signal to the region determination unit 912, and a region determination is performed based on the AC brightness signal input from the color conversion unit 909, and a region determination result is output to the quantization unit 910. At the same time, a region determination unit 912 that outputs a region flag to the encoding unit 911, and a color conversion unit 907 according to the region determination result.
Quantizes the DC brightness signal and DC color signal input from
A second-stage quantization is performed on the AC lightness signal and the AC color signal input from the color conversion unit 909, and the quantization unit 910 outputs the result to the encoding unit 911.
And a coding unit 911 that creates a fixed-length code by combining the information (units of 2 × 2 pixel blocks) quantized by the above and stores it in the page memory 403.

Next, the encoding method of the encoder 402 shown in FIG. 23 will be described in detail. Image data composed of R, G, and B color components (for example, 2 × 2 pixel block units, each pixel of 8 bits) read from the scanner 401 is an n × m buffer 901, 90 of R, G, and B colors.
2 and 903 and are temporarily stored.

The R, G, and B color position correlation conversion unit 904,
905 and 906 are n × m buffers 90 of R, G and B colors
1, 902, and 903 (R
GB) to perform a position correlation conversion to obtain a DC component L for each color.
LR, LLG, LLB and AC components HR, HG, HB are separated, and DC components LLR, LLG, LLB are converted into a color conversion unit 90.
7, while the AC components HR, HG, HB are output to an AC component quantization unit 908. Here, as the position correlation conversion, the Haar transform is performed as in the first embodiment.

The AC component quantization section 908 performs the first-stage quantization of the input AC components HR, HG, and HB of each component and outputs the result to the color conversion section 909. In the first-stage quantization in the AC component quantization unit 908, for example, HH is deleted from the AC components HL, LH, and HH.

The color conversion section 909 includes an AC component quantization section 90
8 is subjected to color conversion (brightness / color separation) of the AC component of each component that has been subjected to the first stage quantization and
It generates an AC lightness signal YH and AC color signals UH and VH, outputs the AC lightness signal YH and the AC color signals UH and VH to the quantization unit 910, and outputs the AC lightness signal YH to the area determination unit.

The color conversion unit 907 determines whether the R, G, B
DC components LLR, LLG, LLB of each component input from the color position correlation conversion units 904, 905, 906
Is subjected to color conversion (lightness / color separation) to generate a DC lightness signal YLL and DC color signals ULL and VLL, and a quantization unit 910.
Output to

The area determining section 912 makes an area determination using the AC lightness signal YH input from the color conversion section 909. Specifically, in the area determination, when the amplitude of YH is equal to or larger than a predetermined value, the 2 × 2 block is determined to be an edge area, and otherwise, a non-edge area. Here, the edge area is an area in which the density change in the block is drastic. In such an area, it is more important to save the resolution than to save the average gradation in the block. On the other hand, a non-edge area is an area in which gradation change in a block is gradual, and in such an area, it is more important to save the average gradation in the block than to save the resolution.

The determination as to whether or not “the amplitude of YH is equal to or more than a predetermined value” is made, specifically, when the absolute value of one of the coefficients YHL and YLH constituting YH is 16 or more, the block is determined. An “edge area” is determined, and other areas are determined as “non-edge areas”. The reason why UH and VH are not used and only YH is used is that YH is a coefficient close to the average value of the AC component and reflects the edges of HR, HG and HB. When area determination is performed using only YH, two conditions “| YHL |> 15” and “| YLH |> are used for the determination.
15 ", the determination can be made easily. Here, | a | is the absolute value of a. Area determination unit 91
2 outputs the region determination result to the quantization unit 910 and outputs the region flag to the encoding unit 911.

The quantizing section 910 receives the area judgment result from the area judging section 912 and performs different quantization for the edge portion and the non-edge portion, and the encoding section 911 collectively compresses the quantized bits. Generate a sign. FIG. 25 shows the format of the compression code. Specifically, the edge portion is YL
L, ULL, VLL 5 bits each, YH 4 bits 13 values,
UH and VH are combined into 25 values and 5 bits. However, since the amplitude of the high frequency component is always 16 or more, (YHL, YLH, YHH) = (0,
0, 0) becomes unnecessary, and the vector of the high-frequency component is
It is sufficient to select 16 values as in the following, and there is no unused code, and the efficiency is high.

As a result, the edge area is converted to a 24-bit compression code whose gradation is coarse but whose resolution is relatively maintained. On the other hand, in the non-edge portion, as shown in FIG.
For L and VLL, nonlinear quantization is performed in the same manner as for the ULL and VLL at the edge portion, and the L and VLL are combined into 6 bits.

The non-linear quantization is performed using a table reference method, and different tables are used as the non-linear quantization tables for the edge portion and the non-edge portion. The non-edge portion is a region in which no high-frequency component having an amplitude of 16 or more exists in the block, so that it can be regarded as a substantially flat region, and all high-frequency components are discarded. Finally, the remaining 5 bits are used as an area flag for the decoding side to recognize whether the target block belongs to an edge area or a non-edge area.

More specifically, UH and VH are set in the edge area at the same position as the 5-bit code representing UH and VH in the edge area (for example, as shown in FIG. 25, the lower 5 bits in the 24-bit compression code). A value that is not used in the expressed 25 values is used as an area flag. In this example, “00” is set as the area flag.
000 ", while" 00001 to 11001 "are used as the 25 values representing UH and VH. This makes it possible to determine the area on the decoding side by looking at the lower 5 bits of the compression code. The non-edge area is a 24-bit compression code having a low resolution but relatively maintaining gradation.

As a result, a code in which important information is stored in each of the edge and non-edge areas can be obtained, and compression with less deterioration in image quality can be achieved. 2 regardless of the area
As a result, a 4-bit fixed-length code can be obtained, and a compressed code that is easy to access in memory and that is excellent in editing such as rotation can be obtained. Further, since the transfer of the code is generally performed in units of one or several bytes, the code of the present example, which is often 24 bits = 3 bytes, is easy to control at the time of transfer.

The method of selecting the area flag is as follows.
As in this example, if the code value corresponding to the one with a high appearance frequency and the region flag among the 25 values representing UH and VH are set to a code value having a high correlation, this is advantageous in a case where it is necessary to perform the secondary compression. is there. Specifically, an HDD may be further provided in addition to the page memory depending on the printer in case of outputting a plurality of images of a plurality of pages. With such a configuration, the fixed-length code stored in the page memory
The next compression or fixed length code is stored in the HDD and sequentially read out, so that a plurality of pages of images can be electronically sorted and a plurality of copies can be output.

[0178] Such a configuration includes a RAM that is fast to access but expensive, and an HDD that is slow to access but low in unit cost of memory.
And achieves high-speed and large-volume electronic sorting. In such a configuration, 2
In the case of performing secondary compression, if a fixed-length code of an area separation type is used, the correlation between codes is reduced due to the mixture of areas, and the secondary compression ratio tends to deteriorate. On the other hand, the code value corresponding to the small proportional value having a relatively high appearance frequency in the edge area and the area flag appearing in the non-edge area are set to a code having a high correlation, so that the secondary compression rate is reduced. Can be reduced even a little.

As the area flag, one bit for the flag may be prepared in the compression code and used, or a bit used as a meaningful coefficient in another area as in this example. The unused value may be flagged. That is, the area flag may use a bit dedicated to the flag, or may be configured integrally with a code representing a coefficient value. Here, to configure integrally, for example, in a 3-bit 8-value code, if the code value is 0, the area is A; if the code value is 1, the area is B; if the code value is 2 to 7, the area is C; This refers to a code configuration in which a certain bit string represents a region and also represents a coefficient value, such that each numerical value of 2 to 7 represents a coefficient value. Which one to use depends on how the bit allocation in each area is allocated, but this bit allocation is different for each system because it is affected by the γ characteristics of the developing device and the like. Therefore, assuming the optimal bit distribution for the system,
What is necessary is just to determine the optimal method of embedding the area flag.

When it is desired to reflect the result of the area determination on the quantization in the AC component quantization unit, HR, HG,
The determination may be made using HB. FIG. 26 shows HR, H
It is a block diagram which shows the structure of the encoding part at the time of performing area determination using G and HB. The configuration and function of the encoding unit shown in FIG. 26 are almost the same as those of the encoding unit in FIG.
Only the differences will be described. In FIG. 26, an area determination unit 1012 includes R color position correlation conversion units 1004,
05, 1006, AC components HR, HG, H
B, a 2 × 2 pixel block area (edge area
(Non-edge area) and determines the result of the
At the same time, it also outputs to the AC component quantization unit 1008. AC component quantization section 1008 quantizes the AC component based on the result of the area determination.

The region determination unit 912 of the encoder 402 in FIG. 23 performs the region determination in two stages of the edge region and the non-edge region. However, the region determination is performed in multiple stages (three or more stages). May be. Hereinafter, a case where the area determination unit 912 performs the area determination in multiple stages will be described.

For example, in an image as shown in FIG.
A certain block is a block determined to be a non-edge as shown in FIG. 27B, and a certain block is a block shown in FIG.
Since the block is determined to be an edge as in (C), the two regions are finely mixed. When such a mixture occurs, as is clear from FIGS. 27B and 27C, the density of the halftone portion of the restored image is different from “46” in the non-edge portion and “36” in the edge portion. At the same time, the average density of the entire block also decreases from “23.5” to “20”. As a result, dark portions may be scattered in some of the halftone dots, or roughness may occur. This roughness is particularly noticeable in a color original because it appears with a change in color. In FIG. 27, an example in which color conversion is not taken into account has been described. However, if color conversion is performed and color conversion coefficients are quantized, this difference generally becomes wider.

Therefore, a "low edge region" is provided between the edge region and the non-edge region, and the deterioration is reduced by mixing the regions in multiple stages. The low edge area is an area in which an edge is present in the block but the amplitude of the high frequency component is not so intense. In this example, three areas are determined as follows.

│YHL│> 60 or │YLH│> 60 → edge area │YHL│ <16 and │YLH│ <16 → non-edge area Other → low edge area

FIG. 28 shows the bit allocation (format) of fixed-length codes in each area in this example. FIG. 29 shows an example of the vector table of the edge area and the low edge area. The bit distribution of the non-edge area and the contents of each bit are the same as those shown in FIG. The bit distribution of the edge area is the same as that shown in FIG.
The contents of the YH vector are as shown in FIG. Since the amplitude of the high-frequency coefficient is large in the edge region of this example, the vector component also includes a large value. As shown in FIG. 28, the bit allocation of the fixed-length code in the “low edge area” is an intermediate bit allocation between the edge area and the non-edge area. That is, YLL is 6 bits, and one bit is absorbed by setting it to 3 bits of the YH vector. Here, YLL is reduced to 6 bits by deleting lower 2 bits of 8-bit YLL obtained as a result of color conversion, and YH vector is reduced to 3 bits by quantizing to a vector shown in FIG. I do. In this example, since the edge region and the low edge region are separated and different vector tables are used, there is also an advantage that a vector suitable for each region can be set finely.

In this example, each area is determined by the lower 5 bits of the fixed-length code, and the edge area represents UH and VH.
A value that is not used in five values is used as an area flag. In this example, “00000” is used as the area flag of the non-edge area, and “00001” is used as the area flag of the low-edge area.
On the other hand, as the 25 values representing UH and VH, “00” is used.
010 to 11010 ". According to this, the region on the decoding side can be determined by looking at the lower 5 bits of the compression code. The block configuration of the encoder for performing such three-stage area determination can be the same as that in FIG.

Referring to FIG. 30, it will be described that the image quality deterioration due to the mixture of the regions can be reduced by separating the regions in multiple stages as in this example. Explaining without considering color conversion for the sake of simplicity, the 2 × 2 pixel block shown in FIG. 11 is determined to be a low edge area, so that the LL component is compared with a case where it is determined to be an edge area. Is accurately reproduced, and the density of the halftone portion restored by this is “3
8 ", which is close to the original image, and the average density of the entire block is" 22 ", which is close to the average density of the original image," 23.5 ". As described above, in this example, it is possible to reduce image quality deterioration due to the mixture of regions.

FIG. 39 shows a fifth embodiment of the decoder 404.
FIG. The decoder 404 of the fifth embodiment shown in FIG. 39 is different from the encoder 402 of the fifth embodiment (see FIG. 2).
This is for decoding the compressed code encoded in 3). The decoder 404 shown in FIG.
1, an inverse quantization unit 922, an area determination unit 932, an inverse color conversion unit 923, an inverse color conversion unit 924, and an AC component inverse quantization unit 92
5, an R color position correlation inverse converter 926, a G color position inverse correlation converter 927, a B color position inverse correlation converter 928, a restored R color nxm buffer 929, and a restored G color nx m buffer 93
0, a restored B color nxm buffer 931.

Next, a decoding method of the decoder 404 in FIG. 39 will be described. The code division unit 921 is provided in the page memory 4
With reference to the lower 5 bits of the compression code (see FIG. 25) of the 2 × 2 pixel block stored in No. 03, if the 5 bits are “00000”, the area is a non-edge area; YLL, ULL, VLL, (vector) Y
H, (vector) UH (vector) VH, and outputs the result to the inverse quantization unit 922, and outputs the area flag to the area determination unit 932. The region determination unit 932 determines an edge region / non-edge region based on the region flag input from the code division unit 921, and outputs a region determination result to the inverse quantization unit 922.

The inverse quantization unit 922 performs inverse quantization with reference to the area determination result of the area determination unit 923, and outputs the restored DC lightness signal YLL, DC color signals ULL, and VLL to the inverse color conversion unit 923. On the other hand, the restored AC lightness signal Y
H, and outputs the AC color signals UH and VH to the inverse color conversion unit 924. Specifically, in the edge portion, 3 bits are added to the lower order of 5 bits of YLL to make 8 bits, and ULL, VL
L is restored to 9 bits using a table for performing inverse transform of nonlinear quantization on the encoding side. Here, YL
When three bits are added to the lower part of the L5 bit, simply adding “000” by bit shift lowers the density, so it is desirable to add an intermediate value of “100”. For YH, the vector 13 in FIG. 6 is restored according to the code value. For UH and VH, nu and nv are restored from the 25-value code in the reverse procedure to that at the time of decoding.

On the other hand, with respect to the non-edge portion, one bit is added to the lower part of the YLL 7 bits to make 8 bits,
L and VLL are restored to 9 bits using a table for performing inverse transform of nonlinear quantization on the encoding side.

The inverse color conversion unit 923 performs the inverse conversion of the YUV conversion on the restored DC lightness signal YLL, DC color signals ULL, VLL, and converts the restored LLR, LLG, LLB into the R color position correlation inverse conversion unit 926. , G color position correlation inverse conversion unit 92
7. Output to the B color position correlation inverse transform unit 928, respectively.

The inverse color conversion unit 924 performs the inverse conversion of the YUV conversion on the restored AC lightness signal YH and AC color signals UH and VH, and outputs the restored HR, HG, HB to the AC component inverse quantization unit 925. Output. The AC component inverse quantization unit 925 inversely quantizes the restored HR, HG, and HB (in this case, adds HH = 0 to each color), and performs an R color position correlation inverse transformation unit 92.
6. Output to the G color position correlation inverse transform unit 927 and the B color position correlation inverse transform unit 928, respectively.

R, G, B color position correlation inverse conversion section 92
6, 927 and 928 are restored (LLR, HR),
(LLG, HG) and (LLB, HB) are each subjected to the inverse transform of the Haar wavelet transform to obtain a restored R color and a restored G
The color and the restored B color are stored in the nxm buffers 929, 930 and 931. As described above, restoration is performed in units of 2 × 2 pixel blocks. If a large amount of memory can be used, the processing time can be reduced by performing each inverse conversion in a table reference method.

As described above, according to the decoding section of the fifth embodiment, it is possible to restore the code information coded by the encoder 402 of the fifth embodiment. FIG.
FIG. 27 is a block diagram showing a configuration of a decoding unit for restoring code information encoded by the encoding unit having the configuration of FIG. 26.

(Embodiment 6) FIG.
It is a block diagram showing Embodiment 6 of 2nd. The encoder 402 according to the sixth embodiment is different from the encoder 402 according to the second embodiment.
(FIG. 13) Configuration in which an area determination unit is added (Embodiment 5)
In which the color conversion unit is deleted). Except that the DC component is not separated into a DC lightness signal and a DC chrominance signal, the configuration is the same as that of the fifth embodiment, and a detailed description is omitted.

The encoder 402 shown in FIG. 32 has R, G, and B components (for example, 8 bits per pixel for each color) input from the scanner 401 in n × m units (for example, 2 × 2 pixel block units). ), An n × m buffer 110 of R, G and B colors for temporarily storing image data consisting of
Conversion using position correlation (hereinafter, “position correlation conversion”) is performed for each of the components (RGB) stored in the n × m buffers 1101, 1102, and 1103 for R, G, and B colors. R, G, and B color position correlation converters that separate the DC component and the AC component for each color and output the DC component to the quantization unit 1110, while outputting the AC component to the AC component quantization unit 1108. 1104, 110
5 and 1106.

The encoder 402 shown in FIG.
Color, G, B color position correlation conversion units 1104, 1105, 1
The AC component of each component input from 106 is
An AC component quantization unit 1108 that performs the first-stage quantization and outputs the result to the color conversion unit 1109;
8 is subjected to color conversion (brightness / color separation) of the AC component of each component that has been subjected to the first stage quantization and
A color conversion unit 1 that generates an AC lightness signal and an AC color signal, outputs the AC lightness signal and the AC color signal to the quantization unit 1110, and outputs the AC lightness signal to the area determination unit 1112.
Region determination unit 1112 that performs region determination based on AC brightness signal input from color conversion unit 1109 and outputs a region determination result to quantization unit 1110 and outputs a region flag to encoding unit 1111 According to the region determination result, the R, G, and B color position correlation conversion unit 110
4, 1105, and 1106, quantizes the DC signal input from the color conversion unit 1109, and performs the second-stage quantization on the AC lightness signal and the AC color signal input from the color conversion unit 1109. A quantization unit 1110 that outputs
The information quantized by the quantization unit 1110 (in units of 2 × 2 pixels) is combined into a fixed-length code, and the page memory 403 is created.
And an encoding unit 1111 for storing the information.

According to the encoder 402 of the sixth embodiment,
Since the DC component is quantized without color conversion, a non-linear quantization table is unnecessary not only on the encoding side but also on the decoding side, and the DC component requires only bit shift for both quantization and inverse quantization. The configuration can be simplified.

When it is desired to reflect the result of the area determination in the quantization performed by the AC component quantization unit, HR, HG,
The determination may be made using HB. FIG. 33 shows HR, H
It is a block diagram which shows the structure of the encoding part at the time of performing area determination using G and HB. The configuration and function of the encoding unit shown in FIG. 33 are almost the same as those of the encoding unit in FIG.
Only the differences will be described. In FIG. 33, the area determination unit 1212 includes R color position correlation conversion units 1204 and 1204.
05, 1206, AC components HR, HG, H
B, a 2 × 2 pixel block area (edge area
(Non-edge area) and determines the result of the
At the same time, the signal is also output to an AC component quantization unit 1208. The AC component quantization unit 1208 quantizes the AC component based on the result of the area determination.

FIG. 34 shows Embodiment 6 of the decoder 404.
FIG. A decoder 404 according to the sixth embodiment shown in FIG. 34 is for decoding the compressed code encoded by the encoder 402 according to the sixth embodiment. FIG.
4 includes a code division unit 1121, an inverse quantization unit 1122, an area determination unit 1132, and an inverse color conversion unit 1
124, an AC component inverse quantization section 1125, an R color position correlation inverse conversion section 1126, a G color position correlation inverse conversion section 1127, B
Color position correlation inverse transform unit 1128, restored R color nxm buffer 1129, restored G color nxm buffer 1130, restored B
And a color n × m buffer 1131. FIG.
FIG. 34 is a block diagram illustrating a configuration of a decoding unit for restoring code information coded by the coding unit having the configuration of FIG. 33.

The encoding method and the decoding method of the encoder 402 and the decoder 404 of the above-described first to sixth embodiments can be implemented by using a program prepared in advance by using a personal computer, a workstation, an image forming apparatus, or the like. May be executed by a computer (CPU or the like).
This program is executed by being read from a computer-readable recording medium such as a hard disk, a floppy (registered trademark) disk, a CD-ROM, an MO, and a DVD. Further, this program can be distributed via the recording medium and as a transmission medium via a network such as the Internet.

[0203] The present invention is not limited to the above-described embodiments, and can be carried out with appropriate modifications without departing from the spirit of the invention.

[0204]

As described above, according to the coding apparatus of the first aspect, the first converting means converts the information of each component into an information amount using the one-dimensional or two-dimensional positional correlation of the image. , And the second converting means converts the information after the conversion by the first converting means to obtain a lightness signal and a color signal.
Quantization is performed on the coefficients obtained by the conversion means and / or the second conversion means. Therefore, before the color conversion, the positional correlation of each component is first used, and then the color conversion is performed. This makes it possible to simplify the encoding circuit and the quantization.

According to the coding apparatus of the present invention, the separating means separates the information of each component into a DC component and an AC component for each predetermined unit, and the converting means converts the AC component separated by the separating means. The components are converted into a lightness signal and a color signal to obtain an AC lightness signal and an AC color signal, and the quantizing means quantizes the AC lightness signal and / or the AC color signal obtained by the conversion means. Since the DC components of each color are not subjected to color conversion, the quantization of the DC components can all be performed by a simple bit reduction method, and there is no need to prepare a table or the like for nonlinear quantization. Further, since the DC component is not color-converted, the circuit for DC component color conversion can be reduced.

According to the encoding apparatus of the third aspect, the separating means separates the information of each component into a DC component and an AC component for each predetermined unit, and the AC component quantizing means separates the information by the separating means. Quantized AC component
The converting means converts the DC component separated by the separating means into a brightness signal and a chrominance signal to obtain a DC lightness signal and a DC color signal, and the second converting means is quantized by the first quantizing means. The AC component is converted into a brightness signal and a chrominance signal to obtain an AC brightness signal and an AC chrominance signal, and the quantization means quantizes the AC brightness signal and / or the AC chrominance signal converted by the second conversion means. Therefore, the circuit scale can be reduced without lowering the coding efficiency as compared with the conventional method.

According to the encoding apparatus of the fourth aspect, the separating means separates information of each component into a DC component and an AC component for each predetermined unit, and the converting means converts the DC component separated by the separating means. Since the components are converted into a brightness signal and a color signal to obtain a DC brightness signal and a DC color signal, and the quantizing means quantizes the AC component separated by the separating means,
Since the AC signal separated by the separating means is quantized as it is by the quantizing means, the encoding circuit can be simplified and the quantization can be simplified.

According to the coding apparatus of claim 5, in the invention of claim 2 or claim 3, the quantizing means performs quantization when the AC brightness signal and / or the AC color signal are quantized. , Vector quantization is performed, so that the processing efficiency can be improved in addition to the effect of the second or third aspect of the present invention.

According to the coding apparatus of claim 6, in the invention of claim 2, 3, or 5, the quantization means quantizes the AC brightness signal and the AC chrominance signal. In doing so, the ratio between the AC lightness signal and the AC color signal is quantized, so that the processing efficiency is further improved in addition to the effect of the invention described in claim 2, claim 3, or claim 5. be able to.

[0210] According to the coding apparatus of the seventh aspect, in the invention of the fourth aspect, the quantizing means vector-quantizes the specific color having the largest edge among the AC components, and converts the other color into another color. Quantizes the ratio with the vector of the specific color, so that in addition to the effect of the invention according to claim 4, it is possible to eliminate the disadvantage that edges disappear during compression.

[0211] According to the encoding apparatus of the eighth aspect, in the invention according to any one of the second to seventh aspects, when the quantization means performs vector quantization,
Since the threshold value is set to be larger than the intermediate point of the quantization value, in the invention according to any one of claims 2 to 7, according to any one of claims 2 to 7, In addition to the effects of the invention described above, it is possible to prevent the occurrence of moire at the halftone dot portion.

According to the encoding apparatus of the ninth aspect, in the invention of the third aspect, the AC component quantizing means deletes one coefficient of the AC component or lower-order bits of all the coefficients to perform the quantization. Therefore, in addition to the effect of the third aspect, it is possible to reduce the size of the color conversion circuit.

According to the encoding apparatus of the tenth aspect, in the invention of the ninth aspect, the AC component quantization means deletes the coefficient corresponding to the oblique edge of the image. In addition to the effects of the invention described in item 9,
It is possible to reduce the data amount without deteriorating the image quality.

[0214] According to the encoding apparatus of the eleventh aspect, in the invention according to any one of the second to tenth aspects, the separating means may include only pixels within a predetermined unit which is a conversion unit. , The conversion is performed with reference to the above. In addition to the effect of the invention according to any one of the second to tenth aspects, the number of accesses to the memory can be reduced and the transfer efficiency can be improved. .

[0215] According to the encoding apparatus of the twelfth aspect, in the invention of the eleventh aspect, the separating means includes:
Since the Haar wavelet transform is performed, the transform is facilitated in addition to the effect of the invention described in claim 11.

Further, according to the encoding apparatus of the present invention, the conversion means performs BTC conversion on the information of each component for each predetermined unit, and the color conversion means performs the color conversion on the BTC converted information, and performs the quantization. Since the means is to quantize the information color-converted by the color conversion means, the coding circuit can be simplified.
The quantization can be simplified.

Further, according to the encoding apparatus of the present invention, the first converting means converts the information of each component into an information amount using a one-dimensional or two-dimensional positional correlation of an image for each predetermined unit. The second conversion means converts the information after conversion by the first conversion means, and the second conversion means obtains a lightness signal and a color signal,
The area determining means determines the area of the image for each predetermined unit, and the quantizing means determines a coefficient obtained by the first converting means and / or the second converting means according to the determination result of the area determining means. Since different quantization is performed for each area, and the flag information creation means creates flag information for identifying the area determined by the area determination means, simplification of the encoding circuit and quantization can be simplified. The compression can be performed with less deterioration of the image quality.

Further, according to the coding apparatus of the present invention, the separating means separates the information of each component into a DC component and an AC component for each predetermined unit, and the first quantizing means uses the separating means. Quantize the separated AC component,
The conversion unit converts the AC component quantized by the first quantization unit into a brightness signal and a color signal to obtain an AC brightness signal and an AC color signal, and the area determination unit performs, for each predetermined unit, According to the magnitude of at least one of the AC component, the AC brightness signal, and the AC chrominance signal, it is determined which of the plurality of areas the predetermined unit belongs to, and the second quantizing means determines the AC brightness. The signal and / or the AC color signal are subjected to different quantization for each area according to the determination result of the area determination means, and the flag information creation means creates flag information for identifying the area determined by the area determination means. Therefore, the encoding circuit can be simplified and the quantization can be simplified, the area can be determined accurately, and compression with less image quality deterioration can be performed.

According to the encoding apparatus of the sixteenth aspect, in the invention of the fifteenth aspect, the second quantizing means may include an AC lightness signal or an AC chrominance signal in a region having a smaller amplitude. Is coarsely quantized, so that in addition to the effect of the invention described in claim 15, it is possible to perform compression with less image quality deterioration depending on the region.

[0220] According to the encoding apparatus of claim 17, in the invention according to any one of claims 14 to 16, the area judging means judges an edge area and a non-edge area. Claims 14 to 1
In addition to the effect of the invention described in any one of the above items 6, compression according to the edge region and the non-edge region can be performed.

According to the coding apparatus of the eighteenth aspect, in the invention according to any one of the fourteenth to sixteenth aspects, the area determining means may be arranged in n (n is an integer of 3 or more) steps. Is determined, it is possible to reduce image quality deterioration due to the mixture of regions.

[0222] According to the coding apparatus of claim 19, in the invention according to any one of claims 14 to 18, the flag information is a flag-only bit or a code representing a coefficient value. Since the flag information is configured integrally, the flag information can be simplified in addition to the effect of the invention according to any one of claims 14 to 18.

According to the encoding apparatus of the twentieth aspect, in the invention according to any one of the first to nineteenth aspects, a fixed length code is generated as code information. In addition to the effects of the invention described in any one of the items 1 to 19, the processing efficiency can be improved.

According to the decoding apparatus of the present invention, the inverse quantization means inversely quantizes the code information to restore the AC lightness signal, the AC chrominance signal, the DC lightness signal, and the DC chrominance signal. The first inverse color conversion means reversely converts the AC lightness signal and the AC color signal to restore the AC signal, and the second inverse color conversion means reversely converts the DC lightness signal and the DC color signal to convert the DC signal. The AC inverse quantization means inversely quantizes the AC signal restored by the first inverse color conversion means, and the inverse position correlation conversion means restores the DC signal and AC inverse quantum information restored by the second inverse color conversion means. Since the AC signal that has been inversely quantized by the converting means is inversely correlated, and each component of the color image is restored, the code information can be restored by a simple circuit configuration and a restoring method.

According to the decoding apparatus of the present invention, the inverse quantization means inversely quantizes the code information to restore the AC brightness signal, the AC chrominance signal, and the DC signal, and performs the first inverse quantization. The means reversely converts the AC lightness signal and the AC color signal to restore the AC signal, and the AC inverse quantization means inversely quantizes the AC signal restored by the first inverse color conversion means, and performs inverse position correlation conversion means. Is to restore each component of the color image by inverse position correlation transform of the DC signal and the AC signal inversely quantized by the AC inverse quantization means, so that the code information can be restored by a simple circuit configuration and a restoration method. It becomes possible.

According to the decoding apparatus of the present invention, the inverse quantization means inversely quantizes the code information, and the inverse color conversion means performs inverse color conversion on the information inversely quantized by the inverse quantization means. Since the inverse BTC means restores each component of the color image by performing inverse BTC conversion on the information subjected to the inverse color conversion by the inverse color conversion means, the code information can be restored by a simple circuit configuration and a restoring method. Become.

According to the decoding apparatus of the present invention, the area determining means determines the area to which the code information belongs from the flag information, and the inverse quantizing means determines the code of the predetermined unit according to the determination result of the area determining means. The information is subjected to different inverse quantization for each area to restore the DC lightness signal, the DC color signal, the AC lightness signal, and the AC color signal, and the first inverse color conversion means reversely colors the AC lightness signal and the AC color signal. The second inverse color conversion means performs inverse color conversion of the DC lightness signal and the DC color signal to restore the DC signal, and the AC inverse quantization means performs first inverse color conversion. Means for inversely quantizing the AC signal reconstructed by the means, and the inverse position transforming means performs inverse position correlation transform of the DC signal restored by the second inverse color transforming means and the AC signal inversely quantized by the AC inverse quantizing means. Each component of the color image Since it was decided to original restoration code information can be a simple circuit configuration and restore.

According to the decoding apparatus of the present invention, the area determining means determines the area to which the code information belongs from the flag information, and the inverse quantizing means determines the code of the predetermined unit according to the result of the determination by the area determining means. The information is subjected to different inverse quantization for each area to restore a DC lightness signal, a DC color signal, an AC lightness signal, and an AC color signal, and the first inverse color conversion means performs inverse color conversion on the AC lightness signal and the AC color signal. Transforming the AC signal, and the inverse position conversion means performs inverse position correlation conversion of the DC signal restored by the inverse quantization means and the AC signal restored by the first inverse color conversion means, thereby converting the color image. Since each component is restored, code information can be restored with a simple circuit configuration and a restoring method.

Further, according to the decoding apparatus of the twenty-sixth aspect, in the invention of any one of the twenty-first to twenty-fifth aspects, the inverse quantization means includes an intermediate quantization means for restoring the DC brightness signal. Claim 21 because restoration is performed using values.
In addition to the effects of the invention described in any one of the twenty-fifth to twenty-fifth aspects, a change in concentration can be reduced and the S / N ratio can be improved.

According to the coding method of the twenty-seventh aspect, the information of each component is stored in a one-dimensional or two-dimensional image.
Using the dimensional positional correlation, the information is transformed so that the amount of information is biased, the transformed information is transformed to obtain a brightness signal and a color signal, and the obtained coefficients are quantized. Therefore, it is possible to simplify the encoding process and the quantization.

According to the encoding method of the twenty-eighth aspect, information of each component is separated into a DC component and an AC component for each predetermined unit, and the separated AC component is converted into a brightness signal and a color signal. Since the AC brightness signal and / or the AC color signal are obtained by conversion and the AC brightness signal and / or the AC color signal are quantized, the DC component is not color-converted, so that the process for color conversion of the DC component is reduced. it can.

According to the encoding method of the present invention, information of each component is separated into a DC component and an AC component for each predetermined unit, the separated AC component is quantized, and the separated DC component is quantized. The components are converted into a brightness signal and a color signal to obtain a DC brightness signal and a DC color signal, and the quantized AC component is converted into a brightness signal and a color signal to obtain an AC brightness signal and an AC color signal. Further, since the converted AC lightness signal and / or AC color signal are quantized,
The number of processing steps can be reduced without lowering the coding efficiency as compared with the conventional method.

According to the encoding method of the present invention, information of each component is separated into a DC component and an AC component for each predetermined unit, and the separated DC component is converted into a brightness signal and a color signal. After converting to obtain the DC brightness signal and the DC color signal, and quantizing the separated AC signal,
It is possible to simplify the encoding process and the quantization.

According to the encoding method of the present invention, the information of each component is stored in the BTC for each predetermined unit.
Since the converted and BTC-converted information is color-converted and the color-converted information is quantized, the encoding process can be simplified and the quantization can be simplified.

[0235] According to the encoding method of the present invention, information of each component is determined so that an information amount is biased by using one-dimensional or two-dimensional positional correlation of an image for each predetermined unit. Perform conversion, perform conversion on the converted information, obtain a brightness signal and a color signal, determine the area of the image for each predetermined unit, and, based on the coefficient obtained by the conversion, determine the area according to the area determination result. Since the fourth step of performing different quantization every time and the flag information for identifying the determined area are created, the encoding step can be simplified and the quantization can be simplified. Thus, compression with less deterioration in image quality can be performed.

According to the encoding method of the present invention, the information of each component is separated into a DC component and an AC component for each predetermined unit, and the separated AC component is quantized. The AC component is converted into a brightness signal and a chrominance signal to obtain an AC brightness signal and an AC chrominance signal. For each predetermined unit, the amplitude of at least one of the DC component, the AC brightness signal, and the AC chrominance signal is obtained. According to the size, it is determined which of the plurality of areas the predetermined unit belongs to, and the AC lightness signal and / or the AC color signal are determined.
Since different quantization is performed for each area according to the area determination result and flag information for identifying the determined area is created, it is possible to simplify the encoding process and the quantization. At the same time, it is possible to accurately determine the area, and it is possible to perform compression with little deterioration in image quality.

Further, according to the decoding method of the present invention, the AC information signal, the AC chrominance signal, the DC luminosity signal and the DC chrominance signal are restored by dequantizing the code information, and the AC luminosity signal and the AC chrominance signal are restored. The signal is inverse-color-converted to restore the AC signal, the DC brightness signal and the DC color signal are inverse-color-converted to restore the DC signal, the restored AC signal is inversely quantized, and the restored DC signal and inverse quantum Since the converted AC signal is subjected to inverse positional correlation conversion to restore each component of the color image, it is possible to restore the code information by a simple decoding process and a simple restoration method.

According to the decoding method of the present invention, the code information is inversely quantized to restore the AC lightness signal, the AC color signal, and the DC signal, and the AC lightness signal and the AC color signal are subjected to the inverse color conversion. Since the AC signal is restored, the restored AC signal is inversely quantized, and the restored DC signal and the inversely quantized AC signal are subjected to inverse positional correlation to restore each component of the color image. Code information can be restored by a simple decoding process and a simple restoration method.

According to the decoding method of the present invention, the code information is inversely quantized, the inversely quantized information is subjected to inverse color conversion, and the information subjected to inverse color conversion by the inverse color conversion means is subjected to inverse BT conversion.
Since each component of the color image is restored by performing the C conversion, it is possible to restore the code information by a simple decoding process and a simple restoration method.

According to the decoding method of the present invention, the area to which the code information belongs is determined from the flag information,
According to the area determination result, a predetermined unit of code information is subjected to different inverse quantization for each area to restore a DC lightness signal, a DC color signal, an AC lightness signal, and an AC color signal, and the AC lightness signal and the AC color signal are reconstructed. Invert color conversion to restore the AC signal, DC brightness signal and DC color signal to inverse color conversion to restore the DC signal, inversely quantized the restored AC signal, inverse quantized AC signal and restored Since each component of the color image is restored by performing the inverse position correlation conversion on the DC signal, the code information can be restored by a simple decoding process and a simple restoration method.

According to the decoding method of the present invention, the area to which the code information belongs is determined from the flag information,
In accordance with the result of the region determination, a predetermined unit of code information is subjected to different inverse quantization for each region to generate a DC signal, an AC brightness signal, and an AC color signal, and perform an inverse color conversion on the AC brightness signal and the AC color signal. Since the AC signal is restored, the restored AC signal is inversely quantized, and the restored DC signal and the inversely quantized AC signal are subjected to inverse positional correlation transformation to restore each component of the color image. It is possible to restore the code information by a simple decoding process and a simple restoring method.

Further, according to the recording medium of claim 39, the program recorded on the recording medium is executed by a computer to execute each step of the invention according to any one of claims 27 to 38. Since the present invention is realized, it is possible to realize each step of the invention according to any one of claims 27 to 38 by executing a program recorded on a recording medium.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a signal processing system of a digital color copying machine according to an embodiment.

FIG. 2 is a block diagram illustrating an encoder according to the first embodiment.

FIG. 3 is a diagram illustrating a two-dimensional Haar transform performed in units of 2 × 2 pixel blocks.

FIG. 4 is a diagram showing an expression of the Haar transform.

FIG. 5 is a diagram showing a color conversion equation.

FIG. 6 is a diagram illustrating an example of a vector quantization table.

FIG. 7 is a diagram comparing vector quantization and scalar quantization.

FIG. 8 is a diagram illustrating the concept of quantization of the ratio between an AC lightness signal vector and an AC color signal vector.

FIG. 9 is a diagram illustrating an example of a quantization value of a ratio.

FIG. 10 is a diagram illustrating an example of a fixed-length code obtained in the first embodiment.

FIG. 11 is a diagram comparing the circuit scale of the prior art with the circuit scale in this embodiment.

FIG. 12 is a block diagram illustrating a configuration of a decoder according to the first embodiment.

FIG. 13 is a block diagram showing an encoder according to the second embodiment.

FIG. 14 is a diagram illustrating a concept of nonlinear quantization of a DC component of a color signal.

FIG. 15 is a diagram showing fixed-length codes obtained in the second embodiment.

FIG. 16 is a diagram illustrating a circuit scale according to the second embodiment.

FIG. 17 is a block diagram illustrating a configuration of a decoder according to the second embodiment.

FIG. 18 is a block diagram illustrating an encoder according to a third embodiment.

FIG. 19 is a diagram illustrating fixed-length codes obtained in the third embodiment.

FIG. 20 is a diagram illustrating a circuit scale according to the third embodiment.

FIG. 21 is a block diagram illustrating a configuration of an encoder according to a fourth embodiment.

FIG. 22 is a block diagram illustrating a configuration of a decoder according to Embodiment 4.

FIG. 23 is a block diagram illustrating a configuration of an encoder according to the fifth embodiment.

FIG. 24 is a diagram illustrating an example of a vector table.

FIG. 25 is a diagram illustrating fixed-length codes obtained in the fifth embodiment.

FIG. 26 is a block diagram showing another configuration of the encoder according to the fifth embodiment.

FIG. 27 is an explanatory diagram for describing a case where region determination is performed at another stage.

FIG. 28 is a diagram illustrating an example of a fixed-length code.

FIG. 29 shows an example of a vector table of an edge area and a low edge area.

FIG. 30 is an explanatory diagram for explaining reduction of image quality deterioration due to mixture of regions.

FIG. 31 is a block diagram showing another configuration of the decoder according to the fifth embodiment.

FIG. 32 is a block diagram illustrating a configuration of an encoder according to Embodiment 6.

FIG. 33 is a block diagram showing another configuration of the encoder according to the sixth embodiment.

FIG. 34 is a block diagram illustrating a configuration of a decoder according to the sixth embodiment.

FIG. 35 is a block diagram showing another configuration of the decoder according to the sixth embodiment.

FIG. 36 is a diagram illustrating an example of a color difference low frequency nonlinear quantization table (5 bits).

FIG. 37 is a diagram illustrating an example of a low-frequency color difference nonlinear quantization table (6 bits).

FIG. 38 is a diagram illustrating an example of a color difference low-frequency nonlinear quantization table (7 bits).

FIG. 39 is a block diagram showing a configuration of a decoder according to the fifth embodiment.

FIG. 40 is a diagram illustrating encoding of a color image based on a conventional uniform color space.

FIG. 41 is a diagram illustrating encoding of a conventional color image.

FIG. 42 is a diagram illustrating encoding of a conventional color image.

[Explanation of symbols]

 Reference numeral 401 Scanner 402 Encoder 403 Page memory 404 Decoder 405 Filter / gradation processing unit 406 Editing panel 501 R color n × m buffer 502 G color n × m buffer 503 B color n × m buffer 504 R color position correlation conversion Unit 505 G color position correlation conversion unit 506 B color position correlation conversion unit 507 Color conversion unit 508 AC component quantization unit 509 Color conversion unit 510 Quantization unit 511 Encoding unit 521 Code division unit 522 Inverse quantization unit 523 Inverse color conversion Unit 524 Inverse color conversion unit 525 AC component inverse quantization unit 526 R color position correlation inverse conversion unit 527 G color position correlation inverse conversion unit 528 B color position correlation inverse conversion unit 529 Restored R color n × m buffer 530 Restored G color n × m buffer 531 Restored B color n × m buffer

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification FI FI Theme Court ゛ (Reference) H04N 11/04 H04N 7/133 Z (72) Inventor Yukiko Yamazaki 1-3-6 Nakamagome, Ota-ku, Tokyo No. F-term in Ricoh Co., Ltd. SS28 TA45 TA55 TC02 TC31 TD09 UA02 UA05 UA39 5C077 LL17 MP08 PP27 PP28 PP31 PP32 PP33 PP36 PP47 PP49 PP68 PQ08 PQ12 RR05 RR21 TT06 5C078 BA54 BA57 BA58 CA01 CA27 DA01 DA02 DB00 5J064 AA02 BC13 BC16 BC13

Claims (39)

[Claims] 1. An encoding apparatus for compressing a color image composed of a plurality of components, each of which is a color component constituting a color image, to generate code information, comprising the steps of: A first conversion unit for converting the information amount so as to be biased by using the positional correlation; and a second conversion unit for converting the information converted by the first conversion unit to obtain a brightness signal and a color signal. An encoding apparatus comprising: a transforming unit; and a quantizing unit that quantizes a coefficient obtained by the transformation by the first transforming unit and / or the second transforming unit. 2. An encoding apparatus for compressing a color image composed of a plurality of components, each of which is a color component constituting a color image, to generate code information, comprising the steps of: A separating unit that separates the AC component into an AC component; a converting unit that converts the AC component separated by the separating unit into a brightness signal and a color signal to obtain an AC brightness signal and an AC color signal; And a quantizing means for quantizing the AC lightness signal and / or the AC color signal. 3. An encoding apparatus for compressing a color image composed of a plurality of components, each of which is a color component constituting a color image, to generate code information, comprising the steps of: Separating means for separating the AC component into AC components, AC component quantizing means for quantizing the AC components separated by the separating means, and converting the DC components separated by the separating means into a brightness signal and a color signal. First to obtain DC lightness signal and DC color signal
And an AC component quantized by the first quantizing means,
Second conversion means for converting the lightness signal and the color signal into an AC lightness signal and an AC color signal, and a quantum for quantizing the AC lightness signal and / or the AC color signal converted by the second conversion means. An encoding device for a color image, comprising: encoding means. 4. A color image encoding apparatus for compressing a color image composed of a plurality of components, each of which is a color component constituting a color image, wherein information of each component is converted into a DC component and an AC component for each predetermined unit. Separating means for converting the DC component separated by the separating means into a brightness signal and a color signal to obtain a DC lightness signal and a DC color signal; andthe AC component separated by the separating means. A color image encoding device, comprising: a quantization unit that performs quantization. 5. The encoding apparatus according to claim 2, wherein said quantization means performs vector quantization when quantizing the AC lightness signal and / or the AC chrominance signal. . 6. The apparatus according to claim 2, wherein said quantizing means quantizes a ratio between the AC lightness signal and the AC color signal when quantizing the AC lightness signal and the AC color signal. Or the encoding device according to claim 5. 7. The method according to claim 1, wherein said quantizing means quantizes a vector of a specific color having the largest edge among said AC components, and quantizes a ratio of another color to a vector of the specific color. Item 5. An apparatus for encoding a color image according to Item 4. 8. The apparatus according to claim 2, wherein said quantization means sets a threshold value larger than an intermediate point of the quantization value when performing vector quantization. An encoding device according to claim 1. 9. The encoding apparatus according to claim 3, wherein said AC component quantization means performs quantization by deleting lower bits of one coefficient or all coefficients. 10. The encoding apparatus according to claim 3, wherein said AC component quantization means deletes a coefficient corresponding to an oblique edge of an image. 11. The encoding apparatus according to claim 2, wherein the separation unit performs the conversion with reference to only a pixel in a predetermined unit that is a conversion unit. apparatus. 12. The encoding apparatus according to claim 11, wherein said separating means performs a Haar wavelet transform. 13. A color image encoding apparatus for compressing a color image composed of a plurality of components, each of which is a color component constituting a color image, to generate code information, comprising the steps of: A color comprising: a conversion unit for performing BTC conversion; a color conversion unit for performing color conversion on the BTC-converted information; and a quantization unit for quantizing information that has been color-converted by the color conversion unit. Image encoding device. 14. A color image coding apparatus for generating code information by compressing a color image composed of a plurality of components, each of which is a color component of a color image, comprising the steps of: First conversion means for converting the information amount so as to cause a bias using the one-dimensional or two-dimensional positional correlation of the image; and performing conversion on the information converted by the first conversion means, A second conversion unit that obtains a signal and a color signal; an area determination unit that determines an area of an image for each of the predetermined units; and a conversion unit that obtains a signal and a color signal. Quantizing means for performing different quantization on each of the areas according to the determination result of the area determining means, and flag information for identifying the area determined by the area determining means are created. Encoding device of a color image, characterized in that it includes a flag information generating means, the that. 15. An encoding apparatus for compressing a color image composed of a plurality of components, each of which is a color component constituting a color image, to generate code information, comprising the steps of: Separation means for separating the AC component separated by the separation means, a first quantization means for quantizing the AC component separated by the separation means, and an AC component quantized by the first quantization means.
Conversion means for converting into a brightness signal and a color signal to obtain an AC brightness signal and an AC color signal; and for each of the predetermined units, at least one of the AC component, the AC brightness signal, and the AC color signal Area determining means for determining which of the plurality of areas the predetermined unit belongs to in accordance with the magnitude of the amplitude; and determining the AC lightness signal and / or the AC color signal by the area determining means. A second quantization unit that performs different quantization for each region according to the result; and a flag information creation unit that creates flag information for identifying the region determined by the region determination unit. Encoding device. 16. The apparatus according to claim 1, wherein the second quantizing means coarsely quantizes the AC lightness signal or the AC chrominance signal in a region where the amplitude is smaller.
6. The encoding device according to 5. 17. The encoding apparatus according to claim 14, wherein the area determination unit determines an edge area and a non-edge area. 18. The encoding apparatus according to claim 14, wherein said area determination means determines an area of n stages (n is an integer of 3 or more). 19. The encoding according to claim 14, wherein the flag information is a flag-only bit, or is integrated with a code representing a coefficient value. apparatus. 20. The encoding apparatus according to claim 1, wherein a fixed length code is generated as the code information. 21. A decoding device for decoding code information of a color image composed of a plurality of components, each of which is a color component of a color image, comprising: an AC lightness signal; an AC color signal; An inverse quantization means for restoring a DC lightness signal and a DC color signal; a first inverse color conversion means for inversely transforming the AC lightness signal and the AC color signal to restore an AC signal; Second inverse color conversion means for performing inverse color conversion on a signal and the DC color signal to restore a DC signal, and AC inverse quantization means for inverse quantizing the AC signal restored by the first inverse color conversion means And an inverse position for inversely correlating the DC signal restored by the second inverse color conversion means and the AC signal inversely quantized by the AC inverse quantization means to restore each component of the color image. correlation Decoding apparatus characterized by comprising: a conversion means. 22. A decoding device for decoding code information of a color image composed of a plurality of components, each of which is a color component constituting a color image, wherein the code information is inversely quantized to obtain an AC lightness signal and an AC color signal. And an inverse quantization means for restoring a DC signal; a first inverse color conversion means for inverse color transforming the AC lightness signal and the AC color signal to restore an AC signal; and a first inverse color conversion means. AC inverse quantization means for inversely quantizing the AC signal restored in the above, and each component of the color image by performing inverse position correlation conversion of the DC signal and the AC signal inversely quantized by the AC inverse quantization means. A decoding device, comprising: an inverse position correlation conversion means for restoring. 23. A decoding device for decoding code information of a color image composed of a plurality of components, each of which is a color component of a color image, comprising: an inverse quantization means for inversely quantizing the code information; Inverse color conversion means for performing inverse color conversion on the information inversely quantized by the inverse quantization means; and inverse BTC conversion of the information inversely color converted by the inverse color conversion means to restore each component of the color image. B
A decoding device comprising: TC means. 24. A decoding apparatus for decoding code information of a color image composed of a plurality of components, each of which is a color component constituting a color image, comprising: An inverse quantization means for performing a different inverse quantization of the predetermined unit of code information for each area in accordance with the determination result of the area determination means to restore a DC lightness signal, a DC color signal, an AC lightness signal, and an AC color signal. First inverse color conversion means for performing inverse color conversion on the AC lightness signal and the AC color signal to restore an AC signal; and inverse color conversion on the DC lightness signal and the DC color signal to restore a DC signal. A second inverse color conversion means for converting the AC signal restored by the first inverse color conversion means, an AC inverse quantization means for inversely quantizing the AC signal restored by the first inverse color conversion means, and a DC signal restored by the second inverse color conversion means. Signal and the inverse AC signal quantized by the AC inverse quantizing means are inversely correlated, and inverse position correlation converting means for restoring each component of the color image is provided. apparatus. 25. A decoding device for decoding code information of a color image composed of a plurality of components, each of which is a color component of a color image, comprising: an area determination unit for determining an area to which the code information belongs from flag information; According to the determination result of the area determination means, performs a different inverse quantization of the code information of a predetermined unit for each area, DC signal,
Inverse quantization means for restoring an AC lightness signal and an AC color signal; first inverse color transformation means for effecting an inverse color transformation on the AC lightness signal and the AC color signal to restore an AC signal; and the inverse quantization. DC signal restored by means and said first signal
And a inverse position correlation conversion unit for performing inverse position correlation conversion on the AC signal restored by the inverse color conversion unit to restore each component of the color image. 26. The decoding apparatus according to claim 21, wherein said inverse quantization means restores the DC brightness signal with an intermediate value when restoring the DC brightness signal. 27. An encoding method for compressing a color image composed of a plurality of components, each of which is a color component constituting a color image, to generate code information, wherein the information of each component is one-dimensional or two-dimensional And a second step of converting the information converted in the first step to obtain a lightness signal and a color signal. And a third step of performing quantization on the coefficients obtained by the conversion in the first step and / or the second step. 28. An encoding method for compressing a color image composed of a plurality of components, each of which is a color component constituting a color image, to generate code information, wherein the information of each component is converted into a DC component for each predetermined unit. A first step of separating the separated AC component into a brightness signal and a color signal to obtain an AC brightness signal and an AC color signal; and the AC brightness signal. And / or a third step of quantizing the AC color signal. 29. An encoding method for generating code information by compressing a color image composed of a plurality of components, each of which is a color component constituting a color image, comprising the steps of: A second step of quantizing the separated AC component, a DC step of converting the separated DC component into a brightness signal and a color signal, and a DC brightness signal. A third step of obtaining a DC color signal; a fourth step of converting the quantized AC component into a brightness signal and a color signal to obtain an AC brightness signal and an AC color signal; A fifth step of quantizing the brightness signal and / or the AC chrominance signal. 30. An encoding method for compressing a color image composed of a plurality of components, each of which is a color component constituting a color image, to generate code information, comprising the steps of: A first step of separating the separated DC component into a brightness signal and a color signal to obtain a DC brightness signal and a DC color signal; and the separated AC component. A third step of quantizing the signal. 31. An encoding method for obtaining code information by compressing a color image composed of a plurality of components, each of which is a color component constituting a color image, wherein the information of each component is subjected to BTC conversion for each predetermined unit. A coding method comprising: a first step; a second step of performing color conversion on the BTC-converted information; and a third step of quantizing the color-converted information. 32. An encoding method for compressing a color image composed of a plurality of components, each of which is a color component constituting a color image, to generate code information, comprising the steps of: A first step of performing conversion so as to cause a bias in the amount of information using the one-dimensional or two-dimensional positional correlation of the following; and performing a conversion on the information converted by the first conversion unit to obtain a brightness signal and a color. A second step of obtaining a signal; a third step of determining an image area for each of the predetermined units; and a coefficient obtained by the conversion of the second step and / or the third step. A fourth step of performing different quantization for each area according to the determination result of the area, and a fifth step of creating flag information for identifying the determined area. Method. 33. An encoding method for compressing a color image composed of a plurality of components, each of which is a color component of a color image, to generate code information, comprising: A first step of separating the AC component into a component and an AC component; a second step of quantizing the separated AC component; and converting the quantized AC component into a brightness signal and a color signal to obtain an AC brightness. A third step of obtaining a signal and an AC color signal; and, for each of the predetermined units, the AC component, the AC brightness signal, and the AC color signal according to the magnitude of at least one of the amplitudes of the AC color signal. A fourth step of determining to which of the plurality of regions the unit belongs; and for the AC lightness signal and / or the AC color signal, each of the regions is different according to the determination result of the region. Fifth step and, the determined sixth step and the encoding method of a color image, characterized in that it comprises a creating flag information for identifying the area to be that quantization. 34. A decoding method for decoding code information of a color image composed of a plurality of components, each of which is a color component constituting a color image, wherein the code information is inversely quantized to obtain an AC lightness signal and an AC color signal. A first step of restoring a DC lightness signal and a DC color signal; a second step of performing an inverse color conversion of the AC lightness signal and the AC color signal to restore an AC signal; A third step of performing inverse color conversion on the DC color signal to restore the DC signal, a fourth step of inversely quantizing the restored AC signal, and the restored DC signal and the inversely quantized A fifth step of performing inverse position correlation conversion on the AC signal to restore each component of the color image. 35. A decoding method for decoding code information of a color image composed of a plurality of components, each of which is a color component of a color image, comprising: an AC brightness signal, an AC color signal, And a first step of restoring a DC signal; a second step of inversely converting the AC lightness signal and the AC color signal to restore an AC signal; and a second step of inversely quantizing the restored AC signal. And a fourth step of performing an inverse position correlation transform of the restored DC signal and the inversely quantized AC signal to restore each component of the color image. The decoding method to use. 36. A decoding method for decoding code information of a color image composed of a plurality of components that are color components forming a color image, wherein: a first step of dequantizing the code information; A second step of performing inverse color conversion on the inversely quantized information; and a third step of performing inverse BTC conversion on the information inversely color-converted by the inverse color conversion unit to restore each component of the color image.
And a decoding method. 37. A decoding method for decoding code information of a color image composed of a plurality of components which are color components constituting a color image, wherein a first step of determining an area to which the code information belongs from flag information. And, according to the determination result of the area, performing a different inverse quantization of the code information of the predetermined unit for each area, a DC brightness signal,
A second step of restoring a DC color signal, an AC lightness signal, and an AC color signal; a third step of performing inverse color conversion of the AC lightness signal and the AC color signal to restore an AC signal; And a fourth step of inverse-color-converting the DC color signal to restore a DC signal; a fifth step of inversely quantizing the restored AC signal; and a step of dequantizing the restored AC signal and the restored AC signal. A sixth step of performing inverse positional correlation conversion on the DC signal and restoring each component of the color image. 38. A decoding method for decoding code information of a color image composed of a plurality of components, each color component constituting a color image, wherein a first step of determining an area to which the code information belongs from flag information And a second step of performing a different inverse quantization on the code information of a predetermined unit for each area according to the result of the area determination to generate a DC signal, an AC brightness signal, and an AC color signal; and A third step of performing an inverse color conversion of the AC color signal to restore an AC signal; a fourth step of inversely quantizing the restored AC signal; and a step of inversely quantizing the restored DC signal and inversely quantized. A fifth step of performing inverse positional correlation conversion on the AC signal to restore each component of the color image. 39. A computer-readable recording medium on which a program for executing each step of the invention according to any one of claims 26 to 38 is recorded.