Through color information and face contour information, this paper proposes a new feature extraction method based on color information and face contour. First of all, color regions are segmented by using improved color extraction algorithm, and analyzed in order to find candidate faces. Then edges of these candidate faces are detected, in accordance with edge detection points. They are matched with characteristics of face contour to identify the exact location. Finally, false faces are excluded by using mosaic template. Experimental results show that the algorithm has higher accuracy rate, high detection speed, and can detect faces of a certain point of view.