Research On The Technology Of Face Tracking And Detection Based On Application

In the field of computer vision and pattern recognition, face tracking and detection has been an important research subject. The subject has extensive application prospect in a wide range of fields such as public safety, access control system, customs authentication, video monitoring, intelligent buildings, and human-computer interaction.Face detection is the basis of face tracking. Because the face detection algorithm is characterized by low computational complexity and simplicity, thus, it has been widely utilized. However, one feature based on detecting usually fails to obtain high accuracy in the complex context. The face detection algorithm based on the skin color is simple, but it often be disturbed by the naked body color and other areas which are similar to the color of skin, therefore, the accuracy is comparative low. At the same time, face detection based on the face contour can solve the problem which exists in the skin color detection at the expense of computational complexity. Therefore, the paper presents a modified face detection algorithm that combined the features of skin color with the face contour. Initially, the algorithm is proposed to distinguish the color-related areas from the complex image, and then locate the position of face from the color-related areas through detecting the face contour. The algorithm has fast processing speed and high accuracy.Face tracking is the purpose of face detection. Because Camshift algorithm can solve the problem of keeping the tracking window unchanged which exists in the Mean shift algorithm, it has been widely applied to the face tracking. However, the algorithm doesn’t have the predictive capability, thus it cannot address the problem of tracking the object when it is shadowed. To address the problem, a modified algorithm which is a combination of Kalman and Camshift is proposed. According to the previous position information of the face, Kalman filter is introduced to predict the possible position of face in the current frame, and then Camshift algorithm is put forward to find the actual position of face in the neighborhood of the possible position. Accordingly, the modified algorithm can predict the state of the target and track the target accurately even when it is sheltered at short time. Simulation results show that the algorithm of face detection algorithm based on the features of skin color and the face contour has high accuracy and high speed. And the modified algorithm of face tracking makes great improvement compared with Camshift, it can track the moving face quickly and accurately, and also has preferable robustness for occlusion. All of these show the algorithm of the face tracking and detection is effective and practical.