Applying mixed gaussian skin models to the automatic face detection

How to detect faces in complex color images is still a challenging problem. We propose an automatic face detection algorithm which consists of five stages: mixed skin model, iterative threshold algorithm, morphological operation, region growing and shape and size restraint. Of these stages, the skin model that we established here is based on the mixed Gaussian distributions on two different color spaces, which achieves better results than other methods. Experimental results show that our algorithm can work very well in the color images which are frontal, side and tilted under varying conditions and complex backgrounds