Mapping The Peking Opera Facial Makeup Onto A Human Face In Video Sequences

Peking Opera Facial Make-ups, which is originated in the life, is a special kind ofmakeup method with national characteristics. It is a living art appeared on the stage withthe actors. It has the unique charm and itself has very high aesthetic, culture andapplication value. Along with the value of traditional cultural heritage protection, moreand more attention is paid to the application research of this graphic art’s digitalmodeling. At the same time, face detection and face tracking have not only confined tothe application category of face recognition. They are also widely used in the pictureand video retrieval, video monitoring，facial expression analysis and gender, race, agediscrimination. And the use in digital entertainment is one of the new applications. Richface expressions can express the realness and interesting of the characters and scenes inanimation and game design. Feature point’s identification which can show faceexpression also gets much attention. We take Peking Opera Facial Make-ups as researchobject in this paper. Through the analysis of this traditional art and the combination ofthe facial make-up and real face in video, we do mapping of the digital facial make-up’s.It involves the computer graphics technology, face feature points positioning technology,image deformation technology and so on. Through the research, we can know thistraditional culture more effectively, and promote the development of the facialmake-u’ps application in digital world.The existing researches of Peking Opera Facial Make-ups are based on standardface model and no study about the real face in videos. This paper we present a newmethod which uses image deformation technology for mapping Peking opera facialmakeup onto a human face with time varying expressions in a video sequence. It takesreal face as the object, and the goal is translating the input that is Peking opera facialmakeup and face video to the video where makeup face has given facial makeup style.In this mapping, the organs of the facial makeup, such like eyes, mouth, are alignedwith that of the human face tightly. At the same time, the shape of artistic patterns of thefacial makeup is well preserved. The related programs which are discussed in this thesisinclude: how to get the vector facial makeup and the boundary points of its patterns,how to position the feature points of facial makeup and face in each video frame, andwhich kind of method we can use to do this mapping. In the earlier stage we analyzed the structure of Peking opera facial makeup and thehand-painted process, and then used SVG format ifle to save them in a facial makeuplibrary. Before mapping, we ifrst get the boundary points by subdividing the contourcurve segments of patterns. It will reduce the computation time with this points that canrepresent the facial makeup to do mapping. Second, we define a set of points on theboundary of organs of the facial makeup as the feature points, as well as theircorrespondences on the human face in the first frame of the video. We track the set ofpoints on the human face using active appearance models (AAM) algorithm andcompute the position of forehead points by using homographic matrix to get theircorresponding points on each video frame. For each video frame, with the feature pointsas the control points, we use moving least squares method to map the facial makeuponto the human face frame by frame and get a video that the human face in the video iswith the Peking opera facial makeup. Comparing with the other methods，the organs inthe facial makeup and the human face are more tightly aligned and the shape of thestrokes on the facial makeup is better preserved in our method.