Study On Techniques Of Face Detection And Tracking Based On Skin Color And System Development

The dissertation carried on the investigations of face detection and tracking. The face detection and tracking are the key points in various facial processing algorithms. The field of facial image processing includes face recognition, gesture estimating, facial expression recognition, video monitoring and so on. And almost all of these are related to face detection and face tracking. Based on the collection and analysis of a lot of the domestic and international discourses and research papers on face detection and tracking in recent years, this paper has lucubrated about establishing of automatic face tracking system. According to the previous research on face detection and tracking, a face detection algorithm based on skin tone is completed, and an automatic face tracking system based on improved object tracking algorithm of mean shift is developed here. The system has achieved crucial improvements on some traditional and classical algorithms of face detection and tracking. The primary work of this paper is as following:1. In this paper a face detection algorithm based on skin color is completed. Referring to a lot of face detection algorithm, a kind of face detection method based on the skin tone characteristics is proposed. First the skin regions are segmented by the model of skin tone, then, face candidate area could be obtained by means of edge information and morphological method, and then facial feature could be extracted from facial candidate region, at last the face can be ensured according to the geometric characteristics. Experimental results show that the algorithm is fast response, high accuracy, suitable for pre-treatment of face detection algorithm as automatically face process system.2. In this paper a face tracking algorithm is completed. This algorithm combines the face detection and the object tracking. Face detection algorithm is used to determine the initial tracking regional; and the improved object tracking algorithm of mean shift is used to track face, the algorithm creates a color model by taking 1D histogram from the H channel in HSV space, then converting the incoming video pixels to a back project image, which is the probability distribution of flesh image. The algorithm is simple calculation, fast tracking, and not depending on face gesture and angle.3. By researching, designing and realizing the algorithm, a platform for automatic face tracking system is completed. The system gathered all of the thesis research methods. The system has a user-friendly interface, visibility and can be directly used as a simple application of automatic face tracking system.