The Research Of Face Detection & Tracking In Video

Face Detection and Tracking in the Video is one of the key techniques of the Digital Surveillance System. It has very high value of research and good future of application.An approach of the face detection based on Structure Classifier & SVM Classifier is proposed toward the face detection in color Video. First, we use typical skin-color character and the distributing rules of the organ of the human face to construct a construct classifier, which can remove most face-dislike candidate from the complexity background of the image rapidly. Secondly, support vector machine- classifier is applied to verifying the face candidates abstracted by a structure classifier further. The accurate and real time of this algorithm has been proved by the experimental results.An approach of single face tracking based on Color-Directional Gradient Angular Histogram is proposed. The color and structure features of the image can be present at the same time. The identifying ability is analyzed deeply. Using Color-Directional Gradient Angular Histogram as the character model of tracked face performs better and more stable in the environment where exists a lot of skin-like background. This has been proved by tests.Based on the VC++ on PC, An automatic localization and tracking system for single face is developed. This system can acquire the video from the Video capture Card, process and analysis in real-time, achieve function of face location, single face tracking, kinescope, replay etc. The rate of tracking is more than 15 fpm.