GLOBAL MOTION ESTIMATION IN MODEL-BASED IMAGE-CODING BY TRACKING 3-DIMENSIONAL CONTOUR FEATURE POINTS

Recently a new type of video coding method called model-based image co ding has attracted much attention as a potential candidate for low bit -rate visual communication services, This technique reconstructs the f acial image with a preknown three-dimensional (3-D) human face model a nd its received model motion parameters, The parameters of the head mo tion are mainly divided into two parts: global motion parameters descr ibe the rigid movement of the head, such as rotation and translation, and local motion parameters which deal with the nonrigid movements of facial expressions, such as the opening and closing of the mouth and e yes. In this paper, we propose a new approach which can estimate the h ead global motion more robustly and accurately, Comparing with the exi sting techniques to match only a few key points, here we extract 3-D c ontour feature points and use chamfer distance matching to estimate he ad global motion, This can improve and enhance the contour tracking pe rformance greatly. We also develop another technique called facial nor malization transform, It maps the facial region of the current input f rame back to the normalized pose of the initial frame, Using this tran sform, we can analyze facial expressions at the same orientation and f ixed region, This simplifies the analysis work a lot, Then, we do our encoding by the clip-and-paste method along with adaptive codebook tec hnique. In the following, the coder and decoder system are briefly des cribed, Since we mainly focus the work on the analysis and synthesis o f the facial portion images, background analysis and bitstream coding technique will not be discussed in this paper.