JPH04342396A - Fuzzy picture processing method - Google Patents

Abstract

PURPOSE:To simply and accurately extract the only flesh color areas at a high processing speed by extracting the flesh color areas based on a picture of a CCD camera and a picture of an infrared ray camera. CONSTITUTION:The RGB output of a CCD camera 1 is converted into a color space coordinate value by a picture processor 3. The thermal picture RGB output of an infrared ray camera 2 is converted into a color space coordinate value by a thermal picture processor 4. Whether or not the picture element is in the flesh color area is estimated by the color space coordinate value by the CCD camera 1. When the color space coordinate value by the infrared ray camera 2 is added to this and the picture element is in the body temperature area, this is judged as being in the flesh color area. The flesh color correction by a fuzzy inference is performed on the extracted flesh color area data. The obtained data is returned to the RGB signal and saved in a frame memory in a printer 10.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing method for outputting an image captured by a video camera to a monitor TV or a video printer and reproducing the image.

0002

2. Description of the Related Art Generally, in color output devices such as VTRs, video printers, and color TVs, it is known that if a person's skin tone is reproduced in a desirable manner, the entire image becomes desirable.

For example, in a conventional video camera, CC
The color data (R, G, B) of the image captured at D is converted into color data (L*, Cab*, Hab°) in the color space coordinate system, and the L*, Cab*, Hab When the value of ° is within a predetermined range, it is estimated to be a skin color region, and color correction using fuzzy inference is applied to this region to obtain video output. However, in the above method, it is difficult to distinguish between a peach fruit or skin-colored clothing and a person's skin, and the accuracy of skin color identification is lacking.

[0004]Also, recently, a person recognition method using a neural circuit model has been proposed, but it requires a huge amount of calculations to be performed on a huge amount of image data, and the processing time for recognizing a person is slow. The problem was that it would be huge.

[0005]

[Problems to be Solved by the Invention] In view of the above-mentioned conventional techniques, it is an object of the present invention to provide a method for accurately extracting only the skin color area of a person using a simple and fast processing method. be.

[0006]

[Means for Solving the Problems] The present invention estimates a skin color area from an image captured by a CCD camera, and corresponds to the skin of a person from the estimated skin color area based on an image captured by an infrared camera. The method is to extract only the skin color area of the person and perform color correction on this extracted skin color area of the person. As this color correction, color correction based on fuzzy inference is preferable.

[0007]

[Operation] Color data is obtained from the image obtained by the CCD camera.
A rough skin color area is estimated using only the data, and by adding thermal image data obtained by an infrared camera to this, only the skin color area of the person is extracted.

[0008] Color correction is performed on the skin color area obtained in this way using fuzzy inference or the like to obtain a preferable image.

[0009]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The method of the present invention will be explained in detail below with reference to the drawings.

FIG. 1 shows a schematic diagram of a video input/output device for carrying out the method of the invention. 1 is a CCD camera used in a video camera, etc.; 2 is an infrared camera for obtaining a thermal image; 3 is an image processing device for processing an image signal of a subject captured by the CCD camera; and 4 is an infrared camera for the infrared camera. A thermal image processing device that processes a thermal image signal of a captured object;
The signal (RGB signal) is converted into a color space coordinate system signal (L*, Ca
b*, Hab°); 6 is the converting unit 5;
A skin color estimating unit compares L*, Cab*, and Hab° converted in , with values corresponding to the skin color area to estimate the front lobe sentence of the subject; 7 is the entire skin color portion of the subject obtained by the skin color estimating unit 6; a skin color region extraction unit 8 that extracts a skin color region of a person based on the thermal image of the other thermal image processing device 4;
9 is a fuzzy correction unit that performs color correction using fuzzy inference on the skin color area of the person obtained by the extraction unit 7; 9 is a function that restores the signals of the skin color area of the person and other parts of the subject that have been corrected by the correction unit; 10 is this R
This is a printer that outputs an image of a subject using GB signals.

FIG. 2 is a diagram showing the screen of the CCD camera 1, in which light input from each part of the object is converted into RGB signals using n×m pixels per screen.

The human-sensory image processing method of the present invention in the video input/output device having the above configuration will be explained in accordance with the flowchart of FIG.

First, by inputting a start signal in step S1, the image processing devices 3 and 4 are initialized in step S2.
The upper left pixel in FIG. 2 is set to (i, j) = (1, 1).

Next, in step S3, a comparison is made to see if the i is greater than the n, and if i is less than or equal to n, then in step S4, a comparison is made to see if the j is greater than the m.
When j is less than or equal to m, in step S5 the CCD camera 1
outputs the RGB output (R, G, B) 11 of pixel (1, 1) to the image processing device 3, and in step S6 this RGB
Convert the output to color space coordinates L*, Cab*, Hab°. Further, the infrared camera 2 outputs the thermal image RGB output (R, G, B) 11 of the pixel (1, 1) to the thermal image processing device 4, and in step S6, the RGB output is converted into color space coordinates L*, Convert to Cab*, Hab°.

In step S7, it is estimated from the values of L*, Cab*, and Hab° of the color space coordinates obtained by the CCD camera 1 whether or not the pixel (1, 1) is in the skin color area. The color space coordinates of L*, Cab*, Ha
Taking into account the value of b°, in step S8, if the pixel (1, 1) is in the body temperature region of the person, it is determined that it is in the skin color region of the person. Here, based on the following equation 1, it is determined that the color is skin color if it is within this area.

[0016]

[Math 1]

In step S9, skin color correction is performed using fuzzy inference on the L*, Cab*, and Hab° data of the skin color area of the person extracted as described above, and the obtained data is converted into RGB data in step S10. The signal is restored and stored in a frame memory (not shown) in the printer 10 in step S11.

After this, j is incremented by one and the pixel (
1, 2) is repeated until j becomes m. Furthermore, when j exceeds m in step S4, i is incremented by one and the same operation is performed from pixel (2, 1) until j becomes m. Repeat until. Finally, step S3
When i becomes larger than n, the operation ends in step S12.

Further, if it is determined in step S7 that the color is not a skin color, the process in steps S8 to S10 is not performed and the data is stored in the frame memory as it is, and if it is determined in step S8 that the color is not human body temperature, fuzzy inference is performed. No corrections will be made.

[0020] As described above, each pixel (i, j) in which only the skin color area of the person has been corrected by fuzzy inference is obtained, and by outputting this to the printer 10, the color of the subject is precisely reproduced. This becomes Noh.

[0021]

As described above, the present invention estimates a skin color area from an image captured by a CCD camera, and also estimates a skin color area of a person from the estimated skin color area based on an image captured by an infrared camera. By extracting only the corresponding skin color area and performing color correction on the extracted skin color area of the person, the skin color area of the person can be extracted easily and quickly, and then appropriate correction such as fuzzy inference can be applied. This makes it possible to reproduce the color of the subject based on human senses and print it out.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing a video input/output device implementing the method of the present invention.

FIG. 2 is a plan view showing one image of a CCD camera.

FIG. 3 is an operational flowchart illustrating the method of the present invention.

[Explanation of symbols]

1 CCD camera 2 Infrared camera 6 Skin color area estimation unit 7 Human skin color area extraction unit 8
Correction unit 9 using fuzzy inference Printer

Claims (2)

[Claims] 1. Estimating a skin color area from an image captured by a CCD camera, and extracting only a skin color area corresponding to the skin of a person from the estimated skin color area based on an image captured by an infrared camera, A human-sensory image processing method characterized by performing color correction on the extracted skin color area of the person. 2. The human-sensory image processing method according to claim 1, wherein the color correction is color correction based on fuzzy inference.