Study On The Morphing Of Color Facial Images Based On Improved MR-ASM

The technique of face morphing, which have been widely used in movies making, game and advertise making, is a practical technology in the field of computer animation. The technique of face morphing is an integrated technology of definition and obtaining of facial features, image warping, and image blending. Existing research results mainly focuses on the improvement of the image warping technique, but this paper focuses on many aspects of the face morphing. The main work and achivements of the dissertation can be summarized as follows:1. Usual face morphing techniques are summarized in three aspects. The principle and effect of face morphing algorithms based on feature mesh, feature lines and feature points are discussed.2. For the needs of the automatic positioning of face image feature, this paper has presented an improved Multi-resolution Active Shape Mode (MR-ASM) of the automatic positioning of face image feature. The new algorithm has used Multi-Layers of Adaboost of face detection algorithm and sample-based statistical approach to optimize the initial value and has used features of human face and the edge of the differential information to modify the match structure of some of the key feature points. The improved algorithm has apparently improved the accuracy and the stability.3. For the current morphing technology,the morphing process is not natural enough and the morphing quality is not stable enough. Address all these issues, this paper presents a new color images morphing technique, the morphing process is divided into two phases, and each phase has color adjustment to gray and color saturation adjustment. Experimental show that this approach can produce fluid and natural transition images. Besides, the morphing effect has a very good adaptation.4. This paper has realized a color facial morphing system based on the improved MR-ASM and the new color image morphing technology. Besides, on this basis, this paper has extended it to form a practical face image system.