Multiple faces detection based on chaos-mutation niche genetic algorithms

The traditional methods of multiple human faces detection have the disadvantages of high complexity and low accuracy. A multiple faces detection method based on chaos-mutation niche GA is proposed. This method is applied to detect the faces with complex backgrounds. The faces in the images are described based on the structure of the facial features and the grayscale distribution; Then the chaos-mutation genetic algorithm based on niche is proposed to search the input images simultaneously for potential face regions at different scales by calculating the adaptable values. The experiments show that the algorithm has reached a quite good result that the average detection accuracy is about 86% and the average implementation time for an image (with size of 256 multiply 256 pixels) is less than 1 second on PC of PIII 450.