| Video-based facial animation with detailed appearance texture |
| |
| 引用本文： | WANG Jin,SUN Hanqiu,PENG Qunsheng. Video-based facial animation with detailed appearance texture[J]. 自然科学进展(英文版), 2006, 16(3): 297-306 |
| |
| 作者姓名： | WANG Jin SUN Hanqiu PENG Qunsheng |
| |
| 作者单位： | 1. Department of Computer Science & Technology， Xuzhou Normal University, Xuzhou 221009， China; 2. State Key Laboratory of CAD & CG, Zhejiang University, Zhejiang 310027， China； 3. Department of Computer Science & Engineering, CUHK Shatin, N.T., Hong Kong |
| |
| 基金项目： | Supported by the Major State Basic Research Development Program of China (Grant No. 2002 CB312101)National Natural Science Foundation of China (Grant No.60403038) |
| |
| 摘 要： | Facial shape transformation described by facial animation parameters (FAPs) involves the dynamic movement or deformation of eyes, brows, mouth, and lips, while detailed facial appearance concerns the facial textures such as creases, wrinkles, etc. Video-based facial animation exhibits not only facial shape transformation but also detailed appearance updates. In this paper, a novel algorithm for effectively extracting FAPs from video is proposed. Our system adopts the ICA-enforced direct appearance model (DAM) to track faces from video sequences; and then, FAPs are extracted from every frame of the video based on an extended model of Wincandidate 3.1. Facial appearance details are transformed from each frame by mapping an expression ratio image to the original image. We adopt wavelet to synthesize expressive details by combining the low-frequency signals of the original face and high-frequency signals of the expressive face from each frame of the video. Experimental results show that our proposed algorithm is suitable for reproducing realistic, expressive facial animations.
|
| 关 键 词： | image based rendering facial animation texture transform |
Video-based facial animation with detailed appearance texture |
| |
| WANG Jin,SUN Hanqiu,PENG Qunsheng. Video-based facial animation with detailed appearance texture[J]. Progress in Natural Science, 2006, 16(3): 297-306 |
| |
| Authors: | WANG Jin SUN Hanqiu PENG Qunsheng |
| |
| Abstract: | Facial shape transformation described by facial animation parameters (FAPs) involves the dynamic movement or deformation of eyes, brows, mouth, and lips, while detailed facial appearance concerns the facial textures such as creases, wrinkles, etc. Video-based facial animation exhibits not only facial shape transformation but also detailed appearance updates. In this paper, a novel algorithm for effectively extracting FAPs from video is proposed. Our system adopts the ICA-enforced direct appearance model (DAM) to track faces from video sequences; and then, FAPs are extracted from every frame of the video based on an extended model of Wincandidate 3.1. Facial appearance details are transformed from each frame by mapping an expression ratio image to the original image. We adopt wavelet to synthesize expressive details by combining the low-frequency signals of the original face and high-frequency signals of the expressive face from each frame of the video. Experimental results show that our proposed algorithm is suitable for reproducing realistic, expressive facial animations. |
| |
| Keywords: | image based rendering facial animation texture transform |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
| 点击此处可从《自然科学进展(英文版)》浏览原始摘要信息 |
|
点击此处可从《自然科学进展(英文版)》下载全文 |
