Positioning corners of human mouth based on local gradient operator

Face recognition has widespread applications in monitoring system, public security and home entertainment etc.. However, in practical application, there are many problems needed to be solved in face recognition technology. This paper presents a method to detect and locate accurately facial feature points based on local gradient operator. With Adaboost algorithm, we first detect roughly the mouth area in the face image, and then extract contours of mouth using local gradient operator. Finally, we use Ostu threshold to extract the binary contour around mouth corners according to the precise location of chain code tracing. Experimental results show that local gradient operator can detect and locate rapidly and accurately human mouth, and it is relatively robust against change of facial expressions as well as noise, which helps to improve face recognition rate.