Human Face Detection And Tracking Algorithm Based On Video Sequences

In recent years, with the development of computer skills and computer vision, intelligent video surveillance system has become a new application direction and a cutting-edge issue of concern. Detection and tracking of moving objects based on image sequences is one of important research topics in the domain of computer vision and digital image processing. In this thesis, with intelligent video surveillance system as the main research background, problems of face detection, face tracking and face images selection were researched, and with the above as base, a relatively complete software system of face detection and tracking was constructed. The main contents of this thesis as follows:(1) Human faces were detected using the information of movement, color and gray scale of face. An improved background modeling method based on YCbCr color space was proposed, and Gabor Feature in human face candidate regions extracted by Gabor Filter were sent to the classifier to decide if the region contained faces or not. This method having integrated use of the information of movement, color and gray scale of human face can eliminate the impacts on detection by light and shadow effectively and increase the accuracy and integrity of detection.(2) Human faces were tracked using the information of color, shape and movement of face. First, the tracking was initialized by the result of the human face detection, the automatic tracking of face was realized; Then, human faces were tracked using Camshift tracking algorithm based on the skin color feature of human face, and a Kalman filter was used to estimate the moving objects’status; Finally, constraints for the shape and motion of face were also joined to the Camshift algorithm according to the disadvantage that the Camshift tracking algorithm is interfered by similar color easily. At last, an automatic face tracking system was realized. This method can resolve the problems appearing in the tracking well;(3) Human face images form face tracking have been selected according to the definition and frontal degree of the face images. In this thesis, according to the actual system needs, a comprehensive evaluation system was designed based on the clarity evaluation function and frontal degree of human face. Experimental proved this system can assess face image effectively.(4) Based on the methods and structure of this dissertation, using VC++development platform and OpenCV algorithm, a simple face detection and tracking system based on video was designed, and through this system verified the feasibility and practicability of the algorithm.In this thesis, human face detection and tracking algorithm has been analyzed and researched, a method of human face detection and tracking with good real-time and interfere rejection was proposed, and a practical automatic face detection and tracking software system was constructed.