Algorithm Study For Face Detection In A Complex Background

On the basis of eigenfaces idea, this paper proposed a new concept as eigen skin-color, and proposed a new method to detect human faces using eigen skin-color. First, RGB color space is transformed into HSI color space. Then, the set of points of skin-color is trained to obtain eigen skin-color. At last, face detection is performed in the skin-color space formed by the eigen skin-color. The experiments show that this method is fast and robust.The eigen skin-color method is for the color images. For gray images, this paper uses face recognition method based on hidden Markov models to detect human faces. First, the set of face images is trained to obtain HMM model of human face. Then, the HMM model is used to detect human faces. The experiments show that this method is effective.In a word, for color and gray images, this paper respectively proposed and realized an effective method to detect human faces.