| 摘要 |
PROBLEM TO BE SOLVED: To provide a method for detecting a skin color in an image with high reliability, even in an image where exposure is excessive or insufficient. SOLUTION: In this method for detecting the skin color in a digital image where pixels exist in an RGB color space, first, a statistical analysis for deciding the mean RGB color values is performed. When the mean value of any color is below a prescribed threshold, nonlinear transformation is applied, in order to move the skin color in the image to a prescribed area in the color space. In the preceding stage, the digital image or a transformed digital image is transformed from the RGB space into a generalized RGB space to generate a gRGB digital image. A skin color pixel is detected in the gRGB digital image, a first skin color image mask is formed on the basis of the detected skin color pixel, a masked gRGB image is generated by using the first skin color image mask, and the skin color image mask is used to find the position of the skin color pixel in a digital color image. |
