Research Of Face Detection And Tracking Algorithm

In recent years, with the development of computer and information technology and requirements of related applications, face detection and tracking and related technology have attracted numerous researchers. Various methods of face detection and tracking appear and related productions come out continuously.This paper reviews the development of face detection and face tracking, introduces some algorithm in detail and analyzes their advantages and disadvantages. The author proposed a new mean shift face tracking algorithm using edge orientation histogram, proposed a method for measuring the similarity between the target and background, proposed a method about how to select characteristics of the target in tracking process in the complex background and developed the face detection and tracking system.In face detection aspect, based on the study and comparison of current face detection algorithms, the author proposed a fast and simple face detection method which uses skin model to attain face candidates and uses the templates to verify the human face.In face tracking aspect, based on the study of current face tracking algorithms, the author proposed a new mean shift face tracking algorithm using edge orientation histogram as tracking character. In the traditional Mean-Shift face tracking algorithm, color histogram is used as the character, but it is not sufficient for describing the face's information, especially when the color of the background is similar to the skin color, this causes error in face tracing. The new method uses edge information and texture information other than color information as characters. Experiments proved that the new algorithm is more accurate than the traditional ones when the color of the background is similar to the color of the face.The author also proposed a method for measuring the similarity between the target and background by calculating the Bhattacharyya factor of PDF. In the tracking process, the character on which target is most dissimilar to the background should be chosen. This method eliminates the impact of the background greatly and has great robustness.Based on the work above, the author developed human face detection and trackingsystem. The system can detect and track face automatically.