Study And Implementation Of Technology Of Face Detection Based On Skin Color Information And Features Of Face Structure

Face recognition has broad application prospect and is an extremely challengeable topic, involving the following popular research area: imagery processing, nerve network, pattern detection and so on. It has the vital role value on identification authentication, criminal distinguished, scene surveillance, image retrieval based on content aspect and so on.A complete face recognition system should include face detection, characteristic withdraws as well as match recognition. Face detection is the first link, also is the essential step which face is distinguished. This article mainly has a thorough research on detection technology of colored face under complicated background. The main research content and the achievement include:(1) It analyzes several quite common kinds of colors space in computer vision domain; based on this, it considers that the difference of person's brightness of skin color is large, therefore it establishes skin color model using YCbCr color space except brightness. separates skin color as well as skin color close background region; uses a best auto-adapted threshold value method to make chart two-value and obtains sketchy face region.(2) In algorithm which examines based on the skin color of face, it introduces some present popular algorithms, like skin color detection and characteristic confirmation algorithm, the skin color examination and division characteristic confirmation method, the skin color examination and Mosiac law confirmation method and so on. In view of only adapting simple background, slow detection speed and low efficiency, it proposes one detection method based on skin color and structure characteristic. Compared with other methods, it has a higher speed, a higher accuracy rate, a better robustness and can carry on face detection effectively.(3) Based on knowledge, it analyzes eye detection algorithm and uses edge projection eye detection algorithm; the method achieves that localization is accurate, speed is fast; on the aspect of edge withdrawal, when Canny operator examination has a better effect using the Canny algorithm to withdraw candidate face region edge; at last it uses the method to judge face region of earlier period whether is face, and sections out the eye, the nose and the mouth position based feature of face structure(4) On the above research foundation, it designs and realizes a user interactive contact surface face detection prototype system.From the experimental result, the method has certain robustness and can realize face detection on complex background, and the speed of examination is quick, the success ratio of detection has achieved 94.5%.