JP2001258049A - Image data conversion method, compression method and its device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a conversion method and a compression method, that can convert and compress received image data into image data with less average information amount using a color appearance model(CAM), while providing appearance in the same color to each image under each visual condition, because the average information amount is high in images which do not take visual condition into account. SOLUTION: An input image data section 1 records image data, and an input visual condition section 2 records visual condition data. A CAM conversion section 3 uses the image data and the visual condition data and a sending amount data section 4 converts image data into a sensing amount and records the sensing amount. A virtual visual condition data section 6 records the visual condition data obtained by a visual condition calculation section 5. A CAM inverse conversion section 7 uses the visual condition data recorded by the sensing amount data and the visual condition data recorded by the virtual vision condition data section 6. An image data recording section 8 records the image data obtained by the CAM inverse conversion section, and an image data compression section 9 compresses the recorded data.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to conversion and compression of image data.

[0002]

2. Description of the Related Art A color television system is a system in which, on a transmission side, a picture is decomposed into signals of three primary colors and sent out, and the reception side combines the signals to reproduce the original color picture. It is. Since the color image has a large data amount and is difficult to perform various signal processing and information processing as image data with the data amount as it is, the color image is digitized, compressed on the transmission side, and decompressed on the reception side. It has become.

In a color printer, a high-level apparatus sends a color image as a compressed signal, and the printing apparatus expands the signal to reproduce an original color image.

Further, when a color image is digitized,
Since it is possible to easily create an image having a color different from the original color, the user has been actively manipulating the colors to achieve the desired color.

In recent years, the International Commission on Illumination Conference CIE-TC1-34
`` The CIE 1997 Interim Color Appearance Mod
el (Simple Version), CIECAM97s "(April, 1998).

The color appearance model is a calculation model that converts tristimulus values of a certain color into mathematical relative quantities representing human perceptual attributes such as lightness and hue in consideration of the visual condition parameters. For example, as shown in FIG.
Is different from image A, but when viewed under visual condition B, image A
It has the same color appearance as when you see it. Recommended as a provisional single model of this color appearance model is the CIECAM97s discussed in this report.

[0007]

As described above, by using the color appearance model, various images having the same color appearance under various visual conditions can be created.
However, while each image under each visual condition has the same color appearance, the average amount of information about the image depends on the visual condition. Generally, there is a problem that an average information amount becomes large in an image in which a visual condition is not considered.

[0008]

SUMMARY OF THE INVENTION The present invention provides an image data compression method for converting an input image to obtain an output image.
A new visual condition is created for the input image given the observation visual condition, and the image is converted to an image with the same color appearance under the visual condition and a small amount of average information using the color appearance model. The compressed image is compressed.

The first embodiment will be described. FIG.
1 shows a configuration of a compression apparatus using a color appearance model. Here, the three primary colors of the NTSC system used in the color television system, red green blue R
The image data represented by GB shall be handled. This compression apparatus includes an input image data unit 1 for recording input image data, an input visual condition unit 2 for recording observation visual condition data of an input image, input image data recorded in the input image data unit 1, and input visual conditions. The CAM conversion unit 3 converts the visual sense condition data recorded in the unit 2 into a human sensation amount, and the sensory amount data unit 4 records the human sensation amount derived by the CAM conversion unit 3.

Further, using the visual condition data recorded in the input visual condition unit 2, a visual condition calculating unit 5 for calculating new visual condition data, and the visual condition data calculated by the visual condition calculating unit 5 are recorded. A virtual visual condition data unit 6, a CAM inverse conversion unit 7 for creating image data using a human sense amount recorded in the sense amount data unit 4 and visual condition data recorded in the virtual visual condition data unit 6, a CAM Inversion unit 7
An image data recording unit 8 for recording the image data created in step 1 and an image data compression unit 9 for compressing the image data recorded in the image data recording unit 8.

Each of these units is realized by a personal computer including a central processing unit and a storage unit, and also by a computer program.

The operation of the first embodiment will be described. As shown in FIG. 2, when image data represented by NTSC RGB is input from an interface (not shown), the image data is first recorded in the input image data unit 1. At the same time, when an observation visual condition of the image data is input from an input device (not shown), the input visual condition unit 2
The visual condition data is recorded in the. Here, since the color appearance model CIECAM97s is used, the visual conditions are the luminance of the adaptive visual field L _A 1 and the tristimulus value X for white.
_{w 1, Y w 1, Z} w 1, the relative luminance Y _b 1 to the background, adaptation degree coefficient F1, ambient influences constants c1, color induction factor N _c 1, a lightness contrast factor F _LL 1.

[0013] In CAM conversion unit 3, the visual condition was recorded as the image data recorded in the input image data unit 1 in the input viewing conditions section 2 data _{L A 1, X w 1,} Y w 1, Z w 1, Y _b 1, F1, c1, N _c 1, F
With _LL 1, lightness J1, chroma C1, into a sensory quantity such hue angle h1. This conversion is performed based on the CIECAM97s model. However, in CIECAM97s, CIE-19
31 Since the tristimulus values X, Y, Z of the XYZ color system are input, the RGB of the NTSC system is converted by the following equation (1).

[0014]

(Equation 1)

In the conversion by CIECAM97s, the brightness J
1, chroma C1, hue angle h1, brightness Q, saturation
Although sensations such as s, hue arc H, and colorfulness M can be obtained, here, only the sensation required for the inverse transformation is used.
The sensory amount data unit 4 records the sensory amount, brightness J1, chroma C1, and hue angle h1 obtained by the CAM converter 3. The visual condition calculating unit 5, using the input visual visual condition of the conditional part 2 data _{L A 1, X w 1,} Y w 1, Z w 1, Y b 1, F1, c1, N c 1, F LL 1 ,
Luminance L _A 2 of the adaptation field is a new visual condition data, the tristimulus values X _w 2 with respect to _{white, Y w 2, Z w 2} , the relative luminance Y _b 2 background,
Adaptation degree coefficient F2, ambient influence constant c2, color induction coefficient N
_c 2, Create lightness contrast coefficient _FLL2 .

The above visual conditions are expressed by the following equations (2) to (4).
The condition that satisfies the conditional expressions (cond1) to (cond3) is selected from N (for example, 100) conditions prepared in advance.
The visual condition is created by selecting the condition that maximizes the sum of the left sides of the conditions (cond1) to (cond3).

[0017]

(Equation 2)

Here, b _r0 ,..., B _r45 , b _{g0 ,.
g45, b b0, ···, b} b45 is a weighting coefficient. This weighting factor is the result of a regression analysis using the amount of change in the visual condition as an explanatory variable and the amount of change in the average information amount of each of the RGB as an objective function, and is obtained in advance. The derivation is performed using a large number of data (for example, 30,000).

FIG. 3 shows a flow for creating the data.
You. First, in step 11, the original image (for example, ISO
/ JIS-SCID standard image N1) R, G, B average information
Report E _R', E_G', E_B'Is derived. Next, in step 12,
Observation of each pixel of the original image
Field brightness L_A'(E.g. 318.31 cd / m^Two), Three against white
Stimulus value X_w'(E.g. 95.05), Y_w'(E.g. 100.00), Z_w'(Example
108.88), the relative luminance Y of the background_b'(E.g. 20.00), order
Sensitivity factor F '(for example, 1.0), ambient effect constant c' (for example,
0.59), color induction coefficient N_c'(E.g. 1.0), brightness control
Last coefficient F_LL'Under (e.g. 1.0), change of CIECAM97s
To the lightness J ', the chroma C', and the hue angle h '
You.

In step 13, based on the brightness J ', the chroma C', and the hue angle h 'corresponding to each pixel of the original image, the luminance L _A "of the adaptive visual field as the reference visual condition (for example, 31.83 cd / m ² ), tristimulus values for white X _w "(e.g. 109.85), Y _w " (e.g. 93.
50), Z _w "(for example, 35.58), the relative luminance Y _b " (for example, 3
0.00), adaptation factor F "(e.g. 0.9), ambient influence constant
c "(e.g., 0.525), the color derivation coefficient _Nc " (e.g., 0.8),
Under lightness contrast factor F _LL "(eg 1.0), CIECA
Create a new image using the inverse transform of M97s.

In step 14, an average information amount E _R ", E _G ", E _B "of each of R, G, and B is derived from the created image. The change amount of the average information amount, the change amount of each visual condition parameter (E _R '-E _R ", E _G ' -E _G ",
E _B '-E _B ", L _A ' -L _A ", X _w '-X _w ", Y _w ' -Y _w ", Z _w '-Z _w ", Y _b ' -Y _b ", F '
-F ", c'-c", N _c '-N _c ", F _LL ' -F _LL ") as one data.

A large number of data are created and stored by repeating the work of creating the data by changing the image, the visual observation condition, and the reference condition. For a large number of stored data, derive the maximum value and minimum value of each data element such as the amount of change in average information amount and the amount of change in adaptation luminance, and calculate the maximum value and minimum value for each data element And normalized to between 0 and 1. The i-th data thus normalized is represented as in the following expression (5).

[0023]

(Equation 3)

Then, a weight coefficient is derived using a large number of data obtained by normalization. The weighting factors for the entropy of R, b _r0 ,..., B _r45, are derived by solving the simultaneous equations shown in the following equation (6).

[0025]

(Equation 4)

Here, n is the number of data. b _r0 , ...
・, B _r45 to b _g0 , ..., b _g45 or b _b0 , ..., b _b45
By setting ΔE _Ri to ΔE _Gi or ΔE _Bi , a weight coefficient relating to the entropy of G and B can be similarly calculated.

[0027] The viewing conditions obtained in the visual condition calculating unit 5 data _{L A 2, X w 2,} Y w 2, Z w 2, Y b 2, F2, c2, N c 2, F LL 2, the virtual visual Recorded in condition data section 6. In the CAM inverse conversion unit 7, the sensory amount data obtained in the sensory amount data unit 4,
Lightness J1, chroma C1, hue angle h1, and visual condition data L _A 2, X _w 2, Y recorded in the virtual visual condition data section 6.
_{w 2, Z w 2, Y} b 2, F2, c2, N c 2, F LL 2 performs the inverse transformation of the model is used to derive the R, G, B values of the NTSC system. This inverse transformation is performed based on CIECAM97s, but since the output of CIECAM97s is tristimulus values X, Y, Z of the CIE-1931 XYZ color system, R, G , B is derived.

The inverse transform of the equation (1) is obtained by the following equation (7).

[0029]

(Equation 5)

In the image data recording section 8, the CAM inverse conversion section 7
Record the image data obtained in. Image data compression unit 9
Then, the image data recorded by the image data recording unit 8 is compressed. For example, compression is performed by arithmetic coding whose compression ratio largely depends on the average information amount of an image.

As described above, according to the first embodiment, a certain visual condition is created for an image observed under a certain visual condition, and the visual condition is created using a color appearance model. By creating a corresponding image, it is possible to convert the image into an image in which the average information amount is reduced based on human color perception, and it is possible to increase the compression ratio by compressing the image.

In the extension, by referring to the expanded image under the visual condition in which the image was created, it is possible to obtain human perceptually the same information as the original image observed under the observation condition. Furthermore, the original image can be obtained by using the expanded image and the created visual condition as the input image and the observation visual condition of the color appearance model, and using the observation visual condition of the original image as the reference visual condition.

Next, a second embodiment will be described. FIG.
1 shows a configuration of a compression apparatus using a color appearance model. The compression apparatus includes an input image data unit 10 for recording input image data, an input visual condition unit 11 for recording visual condition data of an input image, a reference visual condition unit 12 for recording reference visual condition data, and an input visual condition unit 11 An information amount calculation unit 14 for comparing the input visual condition data recorded in the reference visual condition data recorded in the reference visual condition unit 12 and estimating a difference in the average information amount of each corresponding image data, an input image data unit Input image data recorded in 10 and input visual condition part
The CAM conversion unit 13 converts the visual condition data recorded in the CAM 11 into an amount of human sensation.

Further, a sensory amount data unit 15 for recording the human sensory amount derived by the CAM conversion unit 13, a reference visual condition data recorded in the reference visual condition unit 12, and a human sensory data recorded in the sensory amount data unit 15. A CAM inverse conversion unit 16 for generating image data using the sensory amount of the image data, an image data recording unit 17 for recording the image data generated by the CAM inverse conversion unit 16, and an image data compression unit 18 for compressing the image data. .

These units are realized by a personal computer having a central processing unit, a storage unit, and the like, and also by a computer program.

The operation of the second embodiment will be described. FIG. 5 shows a flowchart. First, step s2
In step 1, when image data represented by NTSC RGB is input from an interface (not shown), the image data is first recorded in an input image data unit 10. In step s22, when the observation visual condition of the image data is read from an input device (not shown), the input visual condition data is recorded in the input visual condition unit 11. In step s23,
When the reference visual condition is read, the reference visual condition data is recorded in the reference visual condition unit 12.

Here, the type of visual condition data used is the same as that of the first embodiment because CIECAM97s is used as a color appearance model. In step s24,
The information amount deriving unit 14 estimates the difference between the average information amount of the input image viewed under the input visual condition and the average information amount of the image viewed under the reference visual condition using the respective visual conditions. The estimation is performed by the following equation (8).

[0038]

(Equation 6)

Here, E3 is the average information amount of the input image, E4
The average amount of image information corresponding to the reference visual conditions, L _A 3 are luminance of the adapting field in the input viewing conditions, L _A 4 is the adaptation field in the reference viewing conditions _{luminance, X w 3, Y w 3} , Z w 3 tristimulus values for the white in the input visual _{conditions, X w 4, Y w 4} , Z w 4 are tristimulus values for the white in the reference viewing conditions, F3 is adaptation degree coefficient at the input viewing conditions, F4 is adaptation in the input viewing conditions degree coefficient, Y _b 3 are luminance to the background in the input viewing conditions, Y _b 4 is the luminance to the background in the reference viewing conditions, c3 around affected constants at the input viewing conditions, c
4 is the surrounding influence constant in the reference visual condition, N _c 3 is the color induction coefficient in the input visual condition, N _c 4 is the color induction coefficient in the reference visual condition, F _LL 3 is the lightness contrast coefficient in the input visual condition, and F _LL 4 is This is the brightness contrast coefficient under the reference visual condition.

Also, b ₀ ,..., B ₄₅ are weighting factors. These weighting factors are the results of regression analysis using the amount of change in the visual condition as the explanatory variable and the amount of change in the average information amount as the objective function, as in the first embodiment. It is. Here, in order to obtain the estimated value of the average information amount difference of each of R, G, and B, the weight coefficient corresponding to each of R, G, and B is obtained as in the first embodiment.

Estimating the average information difference between R, G, and B,
The sum is defined as the total average information amount difference. In s25, if the entire average information amount difference is not positive, the input image is determined to have a smaller average information amount, and in s29, the input image is compressed by the image data compression unit 18 (for example, arithmetic coding is performed. Do). If the total average information difference is positive,
The image viewed under the reference visual condition is determined to have a smaller average information amount, and the image under the reference visual condition is created and recorded from s26 to s27. The procedure is the same as in the first embodiment,
The difference is that the inverse transformation is performed using the reference visual condition.
In s28, the image data compression unit 18 compresses the image data corresponding to the reference visual condition (for example, performs arithmetic coding).

As described above, according to the second embodiment, when an observation condition and a reference visual condition are given to a certain input image, the input image and the reference visual condition are used by using the observation visual condition and the reference visual condition. By converting the image under the reference visual condition into an image having a smaller average information amount and compressing the image, the compression ratio can be increased. Also, in the expansion, it is possible to determine which image data is compressed by giving an observation visual condition and a reference visual condition.

In the first and second embodiments, CIECAM97s is used as a color appearance model.
However, the present invention can be applied to other color appearance models in which visual conditions are considered. In addition, although arithmetic coding was mentioned as a compression method, other coding methods are also applicable.The NTSC RGB method was used as the color representation method for input image data, but it can be converted to the CIE-1931 XYZ color system. If a new visual condition is created, a condition that satisfies the conditional expression is selected from predetermined visual conditions, but it can be replaced with any creation method as long as the conditional expression is satisfied. Yes, lightness, chroma, and hue angle are used as sensory quantities, but it is also possible to use sensory quantities such as brightness instead of lightness, colorfulness instead of chroma, and hue arc instead of hue angle.

[0044]

According to the present invention, a certain visual condition is created for an image observed under a certain visual condition, and the image is converted into an image corresponding to the visual condition using a color appearance model. By converting the image into an image having a reduced average information amount based on human color perception and compressing the image, the compression ratio can be increased.

[Brief description of the drawings]

FIG. 1 is an explanatory diagram illustrating an outline of a color appearance model.

FIG. 2 is a diagram illustrating a configuration of an apparatus according to the first embodiment.

FIG. 3 is a flowchart showing a flow of creating data.

FIG. 4 is a diagram showing a configuration of an apparatus according to the first embodiment.

FIG. 5 is a flowchart illustrating a second embodiment.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Input image part 2 Input visual condition part 3 CAM conversion part 4 Perception amount data part 5 Visual condition calculation part 6 Virtual visual condition data part 7 CAM inverse conversion part 8 Image data recording part 9 Image data compression part

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat ゛ (Reference) H04N 9/77 H04N 1/46 Z F-term (Reference) 5B057 CA01 CB01 CD14 CH11 DA16 DA17 DB06 5C057 AA11 BA01 CE04 DA01 DA03 DA04 DA17 EA01 EA02 EE01 EL01 EM12 FB03 GE08 GF03 GJ00 5C066 AA11 BA01 BA17 BA20 CA21 DC06 DD06 EB01 EE04 GA02 GA05 HA03 HA04 JA02 KD07 KE04 KE17 KE19 5C077 MP08 PP32 PP35 PP43 PP46 PP47 PQ12 MA46Q19

Claims (6)

[Claims] 1. An image data conversion method for converting an input image to obtain an output image, wherein a new visual condition is created for an input image given an observation visual condition, and the visual condition is created by a color appearance model. An image data conversion method comprising converting an image having the same color appearance below and having a smaller average information amount. 2. An image data conversion method in which an input image is converted to obtain an output image, wherein an input image to which an observation visual condition is given is converted to an input image by using the given reference visual condition and observation visual condition. If the image with the same color appearance under the reference condition is estimated to have a small average information amount, and the image with the same color appearance under the reference visual condition is an image with a smaller average information amount, the color appearance An image data conversion method characterized by conversion by a model. 3. An image data compression method for compressing image data using the conversion method according to claim 1. 4. An image data conversion apparatus for converting an input image to obtain an output image, comprising: means for inputting observation visual condition data of the input image; An image data conversion apparatus comprising: means for calculating reference visual condition data of converted image data; and means for converting input image data with the calculated reference visual condition data. 5. The image data conversion device according to claim 3, wherein the input image data is converted by a color appearance model, and then converted back to output image data. 6. An image data compression apparatus for compressing image data using the conversion apparatus according to claim 4.