Mouth Detection Algorithm Based On Face Organs

In recent years,using the physiological characteristics of human beings,especially for facial feature information to auto detection and recognition has become a hot research topic in the field of artificial intelligence and computer image processing. Among them,detection and recognition eyes or mouth is the emphases of research. Those researches can be widely applied to facial expression recognition,3D face reconstruction and so on.As a major component of the face organs,mouth has its unique characteristics in the color and shape.What's more,information of mouth is not easily hiding,it's very useful for feature extraction.So,mouth detection, especially detection in complex background is a challenging but important task, it shares a relatively large potential for development in security monitoring in public.The paper aim at detection mouth with single frame image gained from the complex background and based on the analysis and the review of computer vision and digital image processing,the main research work and results of the achievement performance of the following:First,in-depth study the two physiological characteristics of human mouth:the color characteristics and the shape characteristics.Then proposed a new region detection method based on the traditional mosaic method.The method redesign the face block strategy by dexterously utilize face organs,using trisection image to roughly locate the image.Then confirmed the mouth region by mouth feature,and complete a mouth detection system.Second, design a mouth template based on the result of trisection image,then improved Snake model for the outer position,effectively address the impact by the illumination caused by the edge detection of leakage problems and can quickly close to mouth's real contours.Experiment show that the algorithm can greatly improve the detection range and detection rate of adaptation,but is much simple than traditionally detection method. In addition,it has better detection effect for the mouth area detection of complex background in public places.