JP2002199225A - Coding method, decoding method, coder, decoder and recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a coder that quantizes a color image before color conversion at an optimum quantization rate for each component of the image so as to efficiently compress the image while suppressing the effect on the image quality. SOLUTION: An S conversion section 102 converts RGB values of a color image segmented in the unit of 4 pixels into a DC component and an AC component. An RCT(Reversible Component Transform) section 112 transforms the DC component into a luminance signal and chrominance signals. Quantization sections 103-111 quantize the AC component. In the quantization, a small quantization rate is adopted for the G component in the AC component including many luminance components and a larger quantization rate is employed for the other components R, B and an RCT section 113 transforms the quantized components into a luminance signal. Linking the components subjected to color conversion to each other generates a code of a fixed length.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an encoding method, a decoding method, an encoding device, a decoding device and a recording medium for efficiently compressing color image information, for example, a device driver such as an application program or a printer driver. This is a technique suitable for devices that handle images.

[0002]

2. Description of the Related Art Generally, when transmitting and storing a color image, an original image is compressed by encoding in order to reduce the amount of data. Conventionally, as a color image compression method, first, color conversion is performed to divide image information into lightness signals and color signals, and then, if necessary, further information is obtained by using positional correlation for each of the lightness signals and color signals. The basic processing flow is to bias (that is, reduce entropy) and perform quantization in the final stage.

[0003] As an example of such a compression method, color image data is color-converted in block units, a brightness signal is orthogonally converted to DC and AC components, and then both DC and AC components are quantized and coded. , While the color signal is
Only the representative colors in the block are quantized without performing orthogonal transformation.
There is an encoding device for encoding (see Japanese Patent Application Laid-Open No. 63-9282).

However, the above-mentioned conventional method is not necessarily the most suitable method for encoding a color image, and has a problem in terms of a circuit scale and a processing amount. For example, by not using a positional correlation having a stronger correlation than the correlation between colors in the preceding stage of the processing, the circuit scale of the color conversion circuit originally realized by small-scale hardware is increased, or There is a problem that different quantization tables must be provided for the brightness signal and the color signal by performing quantization after color conversion.

[0005]

As a solution to such a problem, the present applicant first performs conversion using the positional correlation of each component before color conversion, and then performs quantization. We have proposed an encoding and decoding method that simplifies the encoding circuit and simplifies quantization by performing color conversion after quantization.

The conversion using the positional correlation has a greater effect of concentrating more important information than color conversion, that is, a greater effect of separating insignificant information, so that insignificant information can be quantized before color conversion. I do.

Although the quantization before color conversion originally has little effect on the image quality, it has an effect on the image quality when it is desired to further increase the compression ratio or reduce the circuit scale (processing amount) at the subsequent stage. Components (R, G, B for RGB images, C, M, C for CMYK images)
It is desirable to set an optimal quantization rate (quantization degree) for each of (Y, K). Here, the quantization rate means “the degree of reduction of the information amount of the quantized data with respect to the original data”.
When the bit is reduced to 7 bits, the quantization rate is 2
And 8 bits of data are reduced by 3 bits to 5 bits
, The quantization rate is 2 ＾ 3 = 8.

The present invention has been made in view of the above-mentioned problems, and an object of the present invention is to perform quantization before color conversion at an optimum quantization rate for each component, thereby suppressing the influence on image quality. An object of the present invention is to provide an encoding method, a decoding method, an encoding device, a decoding device, and a recording medium for performing efficient compression.

[0009]

According to the present invention, at the time of encoding, a color image is cut out in units of four pixels, and RGB values are converted into a DC component and an AC component. The DC component is RCT-converted into a brightness signal and a color signal, and the AC component is quantized.
Among the AC components, the quantization rate of G including a large number of brightness components is reduced, the quantization rates of the other components R and B are increased, and after quantization, the brightness signal is RCT-converted. Concatenated to generate a fixed length code.

In the present invention, a color image is frequency-converted,
When quantizing high-frequency components in the process of performing color conversion after quantizing high-frequency components, efficient quantization is performed while maintaining image quality by increasing the R and B quantization rates.

When decoding, the code is divided into a DC lightness signal, a chrominance signal and an AC lightness signal, and the DC lightness signal, the chrominance signal and the AC lightness signal are respectively subjected to inverse RCT transform to obtain a DC component. And restore the AC component. Of the AC components, the inverse quantization rate of G including a large number of lightness components is minimized and inversely quantized, and the DC component and the inversely quantized AC component are inversely transformed to decode the R, G, and B values. I do.

[0012]

DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be specifically described below with reference to the drawings.

FIG. 1 shows the configuration of an encoding device according to the present invention. In the figure, 101 is a 2 × 2 pixel cutout unit that cuts out 2 × 2 pixels from a color image composed of a plurality of components, 102 is an S conversion unit that separates image information of 2 × 2 pixels into a DC component and an AC component, 103 to 105 are quantizers for quantizing the R AC component, 106 to 108 are quantizers for quantizing the G AC component, 109 to 111 are quantizers for quantizing the B AC component, and 112 is a quantizer. A reversible component conversion unit (RCT unit), 113 is an addition unit, and 114 is a code generation unit.

In the present embodiment, for example, a Harr Wavelet transform (hereinafter, a Haar transform), which is a type of wavelet transform, is used in two dimensions as a position correlation transform (means for separating into a DC component and an AC component).

In the two-dimensional Haar transform, as shown in FIG.
When the conversion is performed in units of × 2 pixels and the values of the four pixels are a, b, c, and d as shown in FIG. 2, the image information of the 2 × 2 pixel block is LL, HL, LH, and HH. It is converted into four coefficients. Among them, the LL coefficient is a DC component, and HL,
The LH and HH coefficients are AC components. Note that the Haar transform of 2 × 2 pixels is called S transform.

If ad is a value of R of each pixel, Rll, Rhl, Rlh, Rhh are obtained.
Similarly, in the case of the value of G of each pixel, G11 to G11
hh is obtained, and the same applies to B.

In the above example, the case where the Haar transform is used as the position correlation conversion has been described. However, the present invention is not limited to this.
Subband transform as a generic term of orthogonal transform such as DCT transform, Slant transform, and Hadamard transform, and frequency transform, and block coding using the fact that the gradation change is moderate locally in the image density can also be applied. It is.

FIG. 3 is a flowchart of the encoding process according to the present invention. In this embodiment, for simplicity, the number of pixels to be compressed is limited to an even number both vertically and horizontally. 0-255
Image data given as RGB data having a value of (8 bits) is output by a 2 × 2 pixel cutout unit 101.
The data is input to the S conversion unit 102 in units of four pixels (step 2
01). The S conversion unit 102 performs conversion according to the equation shown in FIG. 2, and performs R11, Rhl, Rlh, Rhh, Gll, G
hl, Glh, Ghh, Bll, Bhl, Blh, Bh
Generate 12 coefficients of h (step 202).

Here, the three DC components R11, G11, and B11 are not quantized, and are not quantized.
(i.e., an I.I. Component conversion) unit 112 (1), and other AC components are quantized by the quantization units 103 to 111.
After that, it is input to the addition unit 113 (RCT unit 2).
By setting the quantization target to the AC component, the influence on the image quality can be effectively suppressed.

In this embodiment, in the quantization units 103 to 105, Rhl, Rlh, and Rhh are linearly quantized for 2 bits, and in the quantization units 109 to 111, Bhl, Blh, and Bhh are also linearly quantized for 2 bits (step). 20
3). Ghl, Gl containing more lightness components than R and B
h and Ghh are 1 bi in the quantization units 106 to 108.
Linear quantization for t is performed (step 204).

In the case of an RGB image, a component containing a large number of lightness components is generally G. By minimizing the quantization rate of the component containing the most lightness information, efficient quantization according to the importance of the information is achieved. I do.

That is, the AC component of R and B is divided by 4, the AC component of G is divided by half and quantized, and Rhl ', Rlh', Rhh ', Bhl', B
1h ′, Bhh ′, Ghl ′, Glh ′, Ghh ′, and are input to the adding unit 113 (RCT unit 2).

As described above, the color conversion calculation can be easily executed by setting the quantization rate in accordance with the color conversion formula in consideration of the importance of the information. In this embodiment, RGB is converted to YUV according to the following color conversion formula.

Brightness signal Y = (R + 2G + B) / 4 (division by 4 may be rounded down, rounded up, or calculated with a real number) Color signal U = RG Color signal V = BG

The above conversion is a reversible conversion when the division unit at 4 is rounded down using a floor function or rounded up using a ceiling function. Therefore, in these two cases, RCT (Reversible Component) is used.
ent Transform (reversible component transformation). In the present invention, for simplicity, it is abbreviated as RCT regardless of reversibility. The color conversion is basically irreversible. For example, R, G, B are X, Y, Z, Y,
In the case of conversion into I and Q, since the conversion coefficient takes a real number, when Y, I and Q are inversely converted into R, G and B, the conversion becomes irreversible (lossy).

In the present invention, RCT is considered as a color conversion formula.
When the rounding off or rounding up of the division is ignored, the brightness signal Y = (R + 2G + B) / 4 = R / 4 + G / 2 +
Since B / 4, the quantization rate of R and B is 4 and the quantization rate of G is 2
Then, for the AC component of Y, Yhl = (Rhl + 2Ghl + Bhl) / 4 = Rhl /
4 + Ghl / 2 + Bhl / 4 = Rhl ′ + Ghl ′ + B
hl'Ylh = (Rlh + 2Glh + Blh) / 4 = Rlh /
4 + Glh / 2 + Blh / 4 = Rlh '+ Glh' + B
lh′Yhh = (Rhh + 2Ghh + Bhh) / 4 = Rhh /
4 + Ghh / 2 + Bhh / 4 = Rhh ′ + Ghh ′ + B
As in hh ', RCT can be performed by simple addition (step 205). That is, when the RTC is used for the color conversion, the RCT calculation for the AC component of Y can be easily performed by setting the quantization rate between RGB to a predetermined relationship, and the processing amount of the RTC calculation unit can be reduced. Can be. AC color signals U (hl, lh, hh), V (hl, l
h, hh) are set to 0.

The three direct current components R11, G11, and B11 are input to the RCT unit 112 (1) without being quantized, and output to the direct current brightness signal Y11, the direct current color signals U11, and V11.
(Step 206).

DC lightness signal Y11 = (R11 + 2G11)
+ Bll) / 4 (Division by 4 may be rounded down, rounded up, or calculated with a real number) DC color signal Ull = Rll-Gll DC color signal Vll = Bll-Gll

The code generator 114 generates the DC lightness signal Yl
1, the DC color signal Ull, the DC color signal Vll, and the AC lightness signal (Yhl, Ylh, Yhh) are connected as a fixed-length code to generate a code for four pixels (step 207). Performed for pixels (step 20)
8).

Here, when the RGB original data is 8 bits and S
When the transformation is used, as is apparent from the equation of FIG.
The L component and the LH component take a 9-bit value including positive and negative signs, and the HH component takes a 10-bit value.

In this embodiment, since the quantization ratio is set to 4 for components other than G, all the AC components except Ghh after quantization become 8 bits or less after quantization. Normally, in computer software, the unit of data length (variable type in software) is 8 bits, 16 bits, 32 bits.
In order to take a value such as a bit, if the above-described quantization is not performed, the data length input to the RCT unit is 16 b
It has to take it (short type).

However, by the above-described quantization, the RCT
Most of the data length input to the section is 8 bits (cha
r type) (LL component is all 8bi)
t). As a result, the amount of memory consumed in the software can be reduced, and in the case of a circuit configuration, the number of connections can be reduced and the circuit scale can be reduced.

If necessary, Ghl, G
It is also possible to set the quantization rate of the lh component to 2 and the quantization rate of the Ghh component to 4, and to set all inputs to the RCT unit 113 to 8 bits or less. In this case, the quantization rate of the hh component is the same for each component, but is not equal for the other AC components (the quantization rates of the Ghl and Glh components are 2, and the quantization rates of Rhl and Rlh are 4). The point that "the quantization means makes the quantization rate of the component having the largest lightness component out of the AC components of the component smaller than the quantization rates of the other components" is different. Absent. In this case, a third RCT unit (not shown) is provided, and only Yhh: Yhh = (Rhh + 2Ghh + Bhh) / 4 = Rhh ′
+ Ghh ′ × 2 + Bhh. As described above, when the SCT is used for the positional correlation conversion and the RCT is used for the color conversion, the processing rate of the RCT calculation unit is reduced by setting the quantization rate between RGB to a predetermined value for each coefficient of the S conversion. can do.

In the above-mentioned embodiment, only the LL with high importance of information is determined for the U and V components.
The other AC component is input to (1),
(2), the RCT unit 113 (2) can be configured more simply. Of course, similarly to Y, an AC component may be obtained for U and V. For example, when Uhl is obtained for the U component, the input data length can be reduced, but Uhl = Rhl−Ghl = Rhl ′ × 4. -Ghl'x
2 needs to be performed. If this processing is performed by a circuit (multiplying by 4 or 2 means that the connection destination is
It's easy and there is no problem at all, but there are some aspects that are somewhat bypassed when using software. Therefore, in the case of executing by software, it is possible to further simplify the calculation of RCT by adopting a configuration in which only the LL component with high importance of information is obtained (this is Ul
Uhl and Ul, which are relatively less important than l and Vll
h, Uhh, Vhl, Vlh, and Vhh are approximated to 0).

The coefficients of each of the YUVs are input to a code generation unit 114, and are concatenated as fixed-length codes in a predetermined order to form a final code.
l, Yhl, Ylh, Yhh, Ull, Vll bit
The length is 8, 8, 8, 9, 9, 9
9, a fixed-length code of 51 bits in total is generated (FIG. 4).

The image before compression is 4 pixels × 3 colors × 8 bits =
Since it is 96 bits, the image is compressed to about 1/2 in this embodiment.

If necessary, the code generation unit 114 can perform quantization according to the importance of YUV, and then generate a code.

FIG. 5 shows the configuration of the decoding apparatus of the present invention. In the figure, reference numeral 301 denotes a code division unit, and 302 denotes an inverse RT.
C parts (1) and 303 are reverse RTC parts (2) and 304 to 31
2 is an inverse quantization unit, 313 is an inverse S transform unit, 314 is 2 × 2
This is a pixel restoration unit.

FIG. 6 is a flowchart of the decoding process according to the present invention. In this embodiment, for simplicity, the number of pixels to be expanded is limited to an even number both vertically and horizontally.

As the inverse position correlation transformation (means for mixing the DC component and the AC component) in the decoding process, for example, an inverse transformation of Harr Wavelet transformation (hereinafter, Haar transformation), which is a kind of wavelet transformation, is used in two dimensions. . The inverse S-transform of the two-dimensional Haar transform is expressed by using the equation shown in FIG.
This is performed in units of 2 × 2 pixels.

In FIG. 5, when a fixed-length code in which the components of YUV are arranged in the order and the number of bits shown in FIG. 4 is given, the code division unit 301 determines that Yll, Yhl, Yl
h, Yhh, Ull and Vll are taken out and Yll, Ul
1 and Vll are input to the inverse RCT section 302 (1), and Yh
l, Ylh, and Yhh are input to the inverse RCT unit 303 (2) (step 401).

In the inverse RCT section 302 (1), inverse conversion is performed by the normal conversion formula Gll = Yll- (Ull + Vll) / 4 Rll = Ull + Gll Bll = Vll + Gll (step 402).

On the other hand, in the inverse RCT section 303 (2), as described above, since the quantization rate of R and B is 4 and the quantization rate of G is 2, Yhl = Rhl ′ + Ghl ′ + Bhl ′ Uhl = 4Rhl′−2Ghl ′ Vhl = 4Bhl′−2Ghl ′, and the inverse conversion is Rhl ′ = (Uhl + 2Ghl ′) / 4 Ghl ′ = Yhl / 2 + (Uhl + Vhl) '= (Vhl + 2Ghl') / 4.

In this embodiment, as described above, Ul
l, Vll (DC component of the color signal) only, and Uhl,
Since Ulh, Uhh, Vhl, Vlh, and Vhh are approximated to 0, for the AC component, Rhl ′ = Ghl ′ / 2 = Yhl / 4 Ghl ′ = Yhl / 2 Bhl ′ = Ghl ′ / 2 = Yhl / 4 Similarly, Rlh '= Glh' / 2 = Ylh / 4 Glh '= Ylh / 2 Blh' = Glh '/ 2 = Ylh / 4 Rhh' = Ghh '/ 2 = Yhh / 4 Ghh' = Yhh / 2Bhh Inverse RCT can be performed by a simple formula of '= Ghh' / 2 = Yhh / 4 (step 403).

The components subjected to the inverse RCT transformation as described above are inversely quantized by the inverse quantization units 304 to 312 (steps 404 and 405). The inverse quantization rate of the present invention refers to the degree of increase in the amount of information due to inverse quantization,
In the case of this embodiment, the inverse quantization rate of R and B is 4, and the inverse quantization rate of G is 2.

As described above, the influence on the image quality can be effectively suppressed by using the AC component as the object of the inverse quantization. Further, by minimizing the inverse quantization rate of the component containing the most lightness information, efficient inverse quantization according to the importance of the information can be performed. Further, by considering the importance of information and setting the inverse quantization rate according to the color conversion formula, the inverse conversion of the color conversion can be easily performed. When RCT is used for the color conversion, By setting the quantization rate between RGB to a predetermined value, RC
Inverse conversion of T can be easily performed.

In the inverse S conversion unit 313, R
The R, G, and B values for four pixels are calculated from 11 to Bhh (step 406), and the restoration unit 314 calculates the R, G, and B values as 4
The image is restored by inputting it as a pixel value (step 40).
7) The above processing is performed for all pixels (step 408).

In the above configuration example, the inverse RCT section 303
After the inverse transform in (2), inverse quantization is performed.
Without performing the division in the CT unit 303 (2), 4Rhl ′ = Yhl 2Ghl ′ = Yhl 4Bhl ′ = Yhl 4Rlh ′ = Ylh 2Glh ′ = Y1h 4Blh ′ = Y1h 4Rhh ′ = Yhh 2Ghh ′ = Yhh4h With such a configuration, inverse transformation and inverse quantization can be performed simultaneously. Even with this configuration,
"The inverse quantization rate of R and B is 4 and the inverse quantization rate of G is 2" remain unchanged.

As described above, the present invention may be implemented by dedicated hardware, but may be implemented by software using a general-purpose computer system. In the case of implementing by software, a program for executing each step of the processing flowchart described in the embodiment of the present invention is recorded on a recording medium or the like. Performed by

FIG. 7 shows a first configuration example when the present invention is implemented by software. HD via data bus
A D (hard disk), a RAM, and a CPU are connected, and a compression process (encoding process) of an original image (original image) is executed according to the following procedure. The original image recorded on the HDD is read into the RAM according to a command from the CPU. The CPU partially reads the image on the RAM, and performs compression by applying the quantization method of the present invention. The CPU writes the compressed data to another area on the RAM. When all the original images are compressed, the data after compression is recorded on the HDD according to a command from the CPU.

In the same apparatus configuration, a decompression process (decoding process) of a compressed image is executed. The compressed image recorded on the HDD is read into the RAM according to an instruction from the CPU. The CPU partially reads the compressed image on the RAM and expands the image by applying the inverse quantization method of the present invention. The CPU writes the decompressed data to another area on the RAM. When all the compressed images are decompressed, the decompressed data is recorded on the HDD according to a command from the CPU.

FIG. 8 shows a second configuration example when the present invention is implemented by software. HD via data bus
D, RAM1 (in PC), CPU1 (in PC), and printer are connected.

When printing out the original image, the image is compressed, and the compressed data is transmitted to the printer. Since the amount of data to be transmitted to the printer is reduced, the transmission time is shortened, and high-speed printing is possible even in consideration of the time required for compression and decompression. The original image recorded on the HDD is read into the RAM according to a command from the CPU. The CPU 1 partially reads an image on the RAM 1 and performs compression by applying the quantization method of the present invention. The CPU 1 writes the data after compression in another area on the RAM 1. The data after compression is recorded on the RAM 2 in the printer according to an instruction from the CPU 1. The CPU 2 in the printer reads the compressed data and expands the image by applying the inverse quantization method of the present invention. The CPU 2 writes the decompressed data on the RAM 2. After all the data is expanded, the printer prints out the expanded data in a predetermined procedure.

In the above configuration example, the original image is prepared on the hard disk. However, the original image of the present invention is not limited to this. For example, image data read from a scanner or the like or image data captured via a network may be used. Good. In the configuration example of FIG. 8, the compressed image may be output to another external device (such as a printer) via a network.

[0055]

As described above, according to the present invention, the following effects can be obtained.

In the encoding process, (1) since the quantization is performed at the optimum quantization rate for each component, it is possible to perform efficient compression while suppressing the influence on the image quality. (2) The effect on the image quality can be effectively suppressed by using the AC component as the object of quantization. (3) By minimizing the quantization rate of the component containing the most lightness information, efficient quantization according to the importance of the information can be performed. (4) The color conversion calculation can be easily performed by setting the quantization rate according to the color conversion equation. (5) When RCT is used for color conversion, the quantization rate between RGB becomes a predetermined value, and RCT calculation can be easily performed. (6) When S-transformation is used for positional correlation conversion and RCT is used for color conversion, the quantization rate between RGB becomes a predetermined value for each coefficient of S-transformation, and the processing amount of the RCT calculation unit can be reduced. . (7) Since only LL coefficients with high importance of information are generated for color information, the RCT unit can be simplified.

In the decoding process: (8) Since the inverse quantization is performed at the optimal inverse quantization rate for each component, the influence on the image quality can be suppressed. (9) The influence on the image quality can be effectively suppressed by performing the inverse quantization at an optimal inverse quantization rate for each component with the AC component being the target of inverse quantization. (10) Efficient inverse quantization according to the importance of information can be performed by minimizing the inverse quantization rate of the component containing the largest amount of brightness information. (11) Since the inverse quantization rate is set in accordance with the color conversion formula in consideration of the importance of the information, the inverse conversion of the color conversion can be easily executed. (12) When RCT is used for color conversion, the quantization rate between RGB becomes a predetermined value, and inverse conversion of RCT can be easily performed. (13) Since only the LL coefficient with high information importance is obtained, the inverse RCT calculation can be simplified.

[Brief description of the drawings]

FIG. 1 shows a configuration of an encoding device according to the present invention.

FIG. 2 is a diagram illustrating S conversion and inverse S conversion.

FIG. 3 is a flowchart of an encoding process according to the present invention.

FIG. 4 shows a format of a fixed-length code encoded according to the present invention.

FIG. 5 shows a configuration of a decoding device of the present invention.

FIG. 6 is a flowchart of a decoding process according to the present invention.

FIG. 7 shows a first configuration example when the present invention is implemented by software.

FIG. 8 shows a second configuration example when the present invention is implemented by software.

[Explanation of symbols]

 101 2 × 2 pixel cutout unit 102 S conversion unit 103 to 111 quantization unit 112 RCT unit 113 addition unit 114 code generation unit 301 code division unit 302 inverse RCT unit (1) 303 inverse RCT unit (2) 304 to 312 inverse quantum Conversion unit 313 inverse S conversion unit 314 2 × 2 pixel restoration unit

Continued on the front page F-term (reference) 5C057 AA07 CA01 DA06 EA01 EA06 EK04 EM08 EM16 FB03 GJ01 GL02 5C059 KK00 KK08 MA22 MA23 MA24 MA41 MC11 MC32 MC34 ME13 PP15 PP16 SS06 SS20 SS28 UA02 UA05 UA29 5C078A01 DA09 A01 BA16 BC08 BC16 BD03

Claims (24)

[Claims] An encoding method for compressing and encoding an image composed of a plurality of components, the method comprising: converting each component into a DC component and an AC component; Converting to a DC color signal; quantizing an AC component of a first component of the plurality of components at a first quantization rate; and converting the first component of the plurality of components. Quantizing an AC component of a second component including a large number of brightness components at a second quantization rate smaller than the first quantization rate; and converting the AC component after the quantization into an AC brightness signal. And a step of generating code information from the DC lightness signal, the DC color signal, and the AC lightness signal. 2. A decoding method for decoding code information of an image composed of a plurality of components, the method comprising: dividing a DC lightness signal, a DC color signal, and an AC lightness signal from the code information; Inverting the brightness signal and the DC color signal into DC components of a plurality of components; inverting the AC brightness signal into AC components of a plurality of components; and converting a first component of the plurality of components. Dequantizing an AC component at a first inverse quantization rate; and converting an AC component of a second component of the plurality of components, which includes a lightness component more than the first component, to the first inverse quantization factor. Dequantizing at a second dequantization rate smaller than the dequantization rate; and converting the DC component and the dequantized AC component into a plurality of components. And decoding to a component. 3. An encoding device for compressing and encoding an image composed of a plurality of components, comprising: first conversion means for converting each component into a DC component and an AC component; A second converter for converting a DC brightness signal into a DC color signal, a first quantizer for quantizing an AC component of a first component among the plurality of components at a first quantization rate, A second quantization unit that quantizes an AC component of a second component among the plurality of components at a second quantization rate smaller than the first quantization rate; and an AC component after the quantization. An encoding apparatus comprising: a third conversion unit that converts into an AC brightness signal; and a code generation unit that generates code information from the DC brightness signal, the DC color signal, and the AC brightness signal. 4. The plurality of components are R, G, B
4. The encoding apparatus according to claim 3, wherein the second component is a component including the largest number of brightness components. 5. The encoding apparatus according to claim 4, wherein the component containing the most lightness component is G. 6. The plurality of components are R, G, B
And the DC components are Rd and G
4. The encoding apparatus according to claim 3, wherein, when d and Bd are used, the second conversion means for converting the DC component into a DC lightness signal and a DC color signal is represented by the following equation. DC lightness signal Yd = (Rd + 2Gd + Bd) / 4 DC color signal U = Rd-Gd DC color signal V = Bd-Gd 7. The plurality of components are R, G, B
The AC component of the first component is Ra, Ba, the AC component of the second component is Ga, and the AC component after quantization is Raq, Ga.
When q and Baq are used, the third conversion means for converting the AC component after quantization into an AC brightness signal is represented by the following equation, and the first quantization rate for quantizing Ra and Ba is: 4. The encoding apparatus according to claim 3, wherein the second quantization rate for quantizing the Ga is twice as large. AC brightness signal = Raq + Gaq + Baq 8. The AC component R of the first component
The code according to claim 7, wherein the first quantization rate for quantizing a and Ba is 4, and the second quantization rate for quantizing the AC component Ga of the second component is 2. Device. 9. The AC component G of the second component
9. The encoding apparatus according to claim 8, wherein the quantization rate of the predetermined coefficient of a is 4. 10. The encoding apparatus according to claim 9, wherein the predetermined coefficient is an HH coefficient. 11. The encoding apparatus according to claim 3, wherein the color signal includes only a specific component having high importance. 12. The encoding apparatus according to claim 11, wherein said specific component is an LL coefficient. 13. The encoding apparatus according to claim 3, wherein said first transform means is a Haar wavelet transform. 14. A decoding device for decoding code information of an image composed of a plurality of components, comprising: means for dividing a DC lightness signal, a DC color signal, and an AC lightness signal from the code information; Means for inversely converting the brightness signal and the DC color signal into DC components of a plurality of components;
Means for inversely converting the AC brightness signal into AC components of a plurality of components, and a first inverse quantum for inversely quantizing an AC component of a first component of the plurality of components at a first inverse quantization rate. And a second inverse quantization means for inversely quantizing an AC component of a second component of the plurality of components at a second inverse quantization rate smaller than the first inverse quantization rate. Decoding means for decoding the DC component and the inversely quantized AC component into a plurality of components. 15. The plurality of components are R, G,
15. The decoding apparatus according to claim 14, wherein the second component is a component that includes the largest number of brightness components. 16. The decoding apparatus according to claim 14, wherein the component containing the most lightness components is G. 17. The method of claim 17, wherein the plurality of components are R, G,
B, and the DC brightness signal is represented by Y
d, when the DC color signals are U and V, the DC lightness signal and the DC color signal are converted into DC components Gd of a plurality of components.
15. The decoding apparatus according to claim 14, wherein the means for performing the inverse conversion to Rd and Bd is represented by the following equation. Gd = Yd- (U + V) / 4 Rd = U + G Bd = V + G 18. The method of claim 18, wherein the plurality of components are R, G,
B, and the AC brightness signal is Y,
Assuming that the AC components of the plurality of components are Ra, Ga, and Ba, the means for inversely converting the AC brightness signal into the AC components of the plurality of components is represented by the following equation. The first inverse quantization rate to be converted is 2 which is the second inverse quantization rate to inversely quantize the AC component Ga.
15. The decoding device according to claim 14, wherein the number is doubled. Ra = Y / 4 Ga = Y / 2 Ba = Y / 4 19. A first inverse quantization rate for dequantizing the AC components Ra and Ba is 4, and a second inverse quantization rate for inversely quantizing the AC component Ga is 2. 19. The encoding device according to claim 18, wherein 20. The decoding apparatus according to claim 14, wherein the color signal includes only a specific component having high importance. 21. The decoding apparatus according to claim 20, wherein said specific component is an LL coefficient. 22. The decoding apparatus according to claim 14, wherein said decoding means is an inverse Haar wavelet transform. 23. A computer-readable recording medium on which a program for causing a computer to realize each step of the encoding method according to claim 1 is recorded. 24. A computer-readable recording medium on which a program for causing a computer to realize each step of the decoding method according to claim 2 is recorded.