Video Of The Human Face Detection And Tracking Technology

Face detection and tracking is an important research aspect in artificial intelligence and computer vision. As a key technology of face information processing, it has a broad application values in many fields such as video surveillance, content-based image retrieval, videoconference, etc.In this thesis, a study on face detection and tracking in video is presented. After a general review of existing schemes on this particular topic, we choose skin color feature as the main feature of human face, due to the real-time requirement of video. By compared different clustering performance in common color space, we construct the mixed skin color model. Experiment shows that by use of the mixed skin color model, the rate of detection lightness has increased.Based on the mixed skin color model, we study two face detection methods, differ-image-based one and evolutionary-agent-based one. The former approach detects face in object moving region based on motion information in a single frame; while the later utilizes a novel and fast approach to mark and cluster skin-like pixel in color image or frame.In order to decide which region really represents a human face, we proposed a face validate method based on template matching. Experiment shows that the approach can exclude non-face region successfully and quickly, and localize human face accurately.Two face tracking methods with different emphasis are studied in this thesis. The first one tracks face in video by detecting face in every frame using evolutionary-agent-based face detection and Kalman motion prediction. The second one applies an non-rigid object tracking methods to face tracking. The central computational module is based on the mean shift iterations and finds the most probable face target position in current frame.