抄録
A novel method to detect human facial regions from a color image with a complex background has been proposed. To detect human facial regions precisely without imposing any constraints, the method employs Genetic Algorithm (GA) and Neural Network. First, the proposed method picks up skin color pixels from an input image. Second, it detects skin color regions using GA based on the idea that human facial regions are skin color. Finally, it judges whether the region is a human face or not one by one using neural network and a simple knowledge processing. We carried out experiments using various 183 color images. It has shown that the detecting accuracy is 95.0 % and the averaged computation time is 0.47sec on a Pentium II 266MHz personal computer.
