摘要
三维扫描仪可以准确获取人脸的几何形状与纹理,但原始的人脸扫描数据仅为一张连续曲面,不符合实际的人脸结构,无法用于人脸动画。针对此问题,提出了一种由三维扫描数据进行人脸建模的方法,将一个具备完整结构的通用人脸模型与扫描数据进行初步适配,再采用细节重建技术恢复特定人脸的表面细节和皮肤纹理。实验表明,该方法建立的三维人脸模型真实感强,结构完整,可生成连续自然的表情动画。
3D face scanner can obtain precise facial shape and texture. However, the original facial scan data is composed of a continuous surface, which is not consistent with the real facial structure. Consequently, the scan data can not be directly applied in facial animation. A method for face modeling from 3D scan data is proposed, in which a generic model with com- plete facial structure was first used to adapt the scan data, and then detail reconstruction was used to recover the surface detail and the skin texture. The experimental result shows that the generated models adapted precisely to the facial scan surfaces with sufficient fidelity, and continuous face animation can be generated.
出处
《现代电子技术》
2011年第21期79-82,共4页
Modern Electronics Technique
基金
国家"863"高技术研究发展计划(2007AA01Z324)
