JPH04167866A - Color picture processing method - Google Patents

Abstract

PURPOSE:To obtain an excellent picture by deciding a color adjustment quantity while taking attributes of an input color picture signal entirely into account, implementing fuzzy inference so as to adjust the color thereby implementing preferred picture correction processing totally in details. CONSTITUTION:A color scanner is used to scan an original picture to obtain an RGB color signal thereby converting the color signal into a color space lightness, saturation, hue polar coordinate color picture signal. To which part of a face the signal data belongs is obtained as the adaptability to a fuzzy set for each color attribute. A membership function to discriminate lips, cheeks and forehead for each lightness, saturation and hue is defined by utilizing the difference from the existing area and the adaptability is obtained by utilizing the function to obtain to which part of the face each picture element belongs for the lightness, saturation and hue. Based on the final color adjustment quantity obtained in this way, a converted color picture signal is corrected to obtain a preferred face color, the signal is converted into a YMC signal as a color print signal. The signal is converted into the RGB signal when it is displayed on a CRT.

Description

DETAILED DESCRIPTION OF THE INVENTION (a) Industrial Application Field The present invention relates to a color image processing method, and in particular to a method that can express skin tones beautifully in color printers, color copying machines, color display devices, etc. Regarding.

(b) Prior Art For example, in a color copying machine, there are cases where it is desired to obtain a color copy image with beautiful skin tones based on the original color image, rather than faithfully copying the inputted color original image. In this case, complexion is particularly important.

In 1989, ``Selective Color Adjustment for Color Image Processing in Color Space,'' Vol. 18, No. 5 of the Institute of Image Electronics Engineers of Japan, proposes color adjustment that takes ambiguity into account, but only hue information and saturation information are processed. Although the weighting process was performed in the target area, only one area was processed.

(C) Problems to be Solved by the Invention According to the above-mentioned prior art, the color of the entire complexion can be adjusted to the same hue or saturation.

However, strictly speaking, the complexion is the sum of the colors of each part of the face, such as lips, cheeks, and forehead, but in order to make the color more beautiful and fine-grained, it is desirable to adjust the color for each part of the face. .

Therefore, the present invention aims to perform preferable facial color adjustment by taking into account all three pieces of information: hue, saturation, and brightness, and performing color adjustment for each part of the face.

(d) Means for Solving the Problems The present invention first converts an input color image signal into a color space L'' (lightness), Cwab' (chroma), and Hlb' (hue) system color image signal. Then, the extent to which each attribute value of H8, Cmb', H3, and " of this converted color image signal matches each predetermined part of the face is determined.

Furthermore, for each part of the face, H8, Cm b', Hab'
A color adjustment amount is determined for each attribute of L*, Cab', and Hlo of the converted color image signal based on a predefined color adjustment amount for each attribute and the degree of conformance determined in the step. , the converted color image signal is corrected based on this color adjustment amount.

Furthermore, the goodness of fit is determined for each part of the face and H3, c lb”
The configuration is such that it is determined as a membership value of a membership function determined for each attribute of IH,,'.

Then, by performing fuzzy inference based on the membership value and the color adjustment amount predefined for each part of the face and each attribute of Ll, Cab', and Hlo, L'' of the converted color image signal, Cab'', H3,.

The configuration is such that the color adjustment amount is determined for each attribute.

(E) Effect According to the above configuration, the three attributes of the input color image signal are L" (lightness), c, 2 (chroma), H3, °
The amount of color adjustment is determined for each part of the face, taking into account all (hue).

Furthermore, by performing fuzzy inference and performing color adjustment,
Image processing that takes three attributes into consideration can be realized with a simple configuration.

(F) Example Hereinafter, a case in which color image processing is performed to obtain a preferable complexion will be explained based on the flowchart shown in FIG. This embodiment is an example of a color copying machine.

First, an RGB color signal is obtained by scanning the original image with a color scanner. This RGB color signal is C
Color space L” (lightness) defined by IE (International Commission on Illumination)
, C-b'' (saturation), and H-b' (hue) are converted into polar coordinate system color image signals.

Next, the converted ", Cab", H1b' signal data (
Each pixel (pixel) belongs to which part of the face, each color = 4
- Find the degree of suitability for the fuzzy set for each attribute, ie, brightness, saturation, and hue.

That is, it is known that the colors of lips, cheeks, and forehead that make up the complexion have different regions in a color space composed of lightness, saturation, and hue. - It is known that when fishing on a boat, the hue of the cheeks is higher in the red direction than the hue of the forehead, and the saturation of the cheeks is slightly higher than the saturation of the forehead. Utilizing such differences in the existence areas, membership functions for distinguishing lips, cheeks, and forehead as shown in FIGS. 2, 3, and 4 are defined for each brightness, saturation, and hue. . Then, by determining the degree of compatibility using this membership function, it is determined to which part of the face each pixel belongs for each brightness, saturation, and hue.

For example, if the values of L'', Cah'', H8, ゛ of a certain pixel are respectively L''1, C, b''l, H1b'l, the degree of suitability of this pixel to the lips is: Lightness L1=μLl(L” l) Saturation C1=μc1 (C1b.1) Hue H1=μmI (Hab' 1 ).

Among these values, the minimum value is taken as the degree of suitability of this pixel to the lips. Let this value be μC1.

Similarly, the suitability for cheeks is: Lightness L2=μm, 2(L Umbrella 1) Saturation C2=μm. r(C, b” 1) Hue H2=μ
It becomes R1 (Hahol).

Among these values, the minimum value is defined as the degree of suitability of this pixel for the cheek. Let this value be μb1.

Similarly, the suitability for the forehead is: Lightness L3=μLl(L'l) Saturation C3" 11 cs(C, b" 1) Hue H
3=μll5cHsb' 1 ).

Among these values, the minimum value is taken as the degree of suitability of this pixel to the forehead. Let this value be μC1.

In this way, the degrees of suitability μal, μb1, and μC1 of one pixel for lips, cheeks, and forehead are determined.

Now, the following table shows the amount of color adjustment in each direction of brightness, saturation, and hue for each of the lips, cheeks, and forehead.

This adjustment amount is a predetermined amount.

Lips Lightness direction 11 Saturation direction C1 Hue direction hl Cheeks Lightness direction j2 Saturation direction C2 Hue direction h2 Forehead Lightness direction 13 Saturation direction C3 Hue direction h3 This predetermined color adjustment amount is corrected based on the degree of suitability. to determine the final color adjustment amount.

That is, the color adjustment amount is determined by using simplified fuzzy inference and weighting by the centroid method.

That is, the color adjustment amount ΔL of the pixel 1 in the brightness direction, the color adjustment amount ΔC of the pixel 1 in the saturation direction, and the color adjustment amount ΔH of the pixel 1 in the hue direction can be determined, respectively.

In this manner, the converted color image signal is corrected based on the determined final color adjustment amount to obtain a desirable complexion.

A signal for color printing is obtained by converting this signal into Y (yellow), M (magenta), and C (cyan) signals and outputting them. When displaying on a CRT display device, it is converted into an RGB signal.

(G) Effects of the Invention According to the present invention described above, for example, image correction processing for obtaining a desirable complexion can be comprehensively and precisely performed, and a good image can be obtained.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing a flowchart of the image processing method according to the present invention, and FIGS. 2, 3, and 4 respectively show brightness,
FIG. 7 is a diagram showing membership functions for determining lips, cheeks, and forehead for each saturation and hue.

Claims (2)

[Claims] (1) A conversion step of converting the input color image signal into color space L^* (lightness), C_a_b^* (chroma), H_a_b' (hue) system color image signals, and L of the converted color image signal. ^*, C_a_b^*, H
a step of determining the extent to which each attribute value of _a_b' matches each predetermined part of the face;
For each attribute of L^*, C_a_b^*, H_a_b' of the converted color image signal, based on the predefined color adjustment amount for each attribute of b' and the degree of suitability determined in the step, The step of calculating the color adjustment amount and the L^*, C_a_b^*,
A color image processing method comprising: a correction step of correcting the converted color image signal based on a color adjustment amount for each attribute of H_a_b'. (2) The degree of fitness is determined for each part of the face and L^*,C
It is determined as the membership value of the membership function defined for each attribute of _a_b^*, H_a_b′,
For each part of the face and L^*, C_a_b^*, H_a_
L^*, C_a_b^*, H_a of the converted color image signal by performing fuzzy inference based on the predefined color adjustment amount and the membership value for each attribute of b'
2. The color image processing method according to claim 1, wherein a color adjustment amount is determined for each attribute of _b'.