Chin Contour Extraction Based on an Auto-Initialized Shape-Enhanced Snake

In this paper, an improved method to extract the chin contour under different face poses is proposed. Several previous approaches have used active contour models(ACM) or snakes in front-view facial images. However, successful chin contour extraction for a wide range of face poses has seldom been mentioned. An algorithm based on an improved snake for chin contour extraction is presented here and it is able to be operated successfully with wide range of different face poses. Since lots of algorithms have been proposed for extracting face region and important facial features like eyes and lips, the algorithm in this paper assumes that the face region and locations of these two features are known as prior knowledge. At the first stage, the face is normalized and face pose is roughly estimated according to the geometric relationship between eyes and lips. Then, the snake for chin contour is initialized with two parabolas according to different face poses. The shape-enhanced snake model is employed to fit the chin contour at last. Experimental results show that the proposed algorithm can extract the chin contour under different face poses with good accuracy.