An improved algorithm of face tracking

This article briefly introduced the main algorithm of face tracking, and put forward an improved algorithm. First, the improved algorithm makes the inter-frame difference processing utilization the information of face´s video-image, and obtains the face region. Then, the algorithm establishes the complexion model and pick up the complexion area. Consequently, the algorithm makes the integral projection to obtain the ellipse area, and obtain the face´s character curve to validate the existing face in the complexion area. At last, used the tensor PCA to extract the feature of human face, and realized the face tracking. The experiment showed that the algorithm can track the human face effectively.