Utilizing Skin Mask and Face Organs Detection for Improving the Viola Face Detection Method

Abstract

According to wide variety of face applications, a fast and reliable face detection approach is one of the most disturbances in the image processing applications. One of the fast and successful approaches in this field was presented by Viola. Low speed, insensitive to faces with partially missing parts and false detection are open problems of this approach. In this paper, skin mask is used to limit the candidate place and scale for detecting face which makes final algorithm runs 4 times faster. Moreover, training this approach to detect eyes and nose makes possible face detecting only with two eyes or one eye and nose, in addition to decrease false detection rate rather than viola approach. Implementing of this approach shows, the presented algorithm could reduce false negative from 10 percent to 2.4 and reduce false positive from 4.4 percent to 2.4.