基于关键特征点的人脸纹理映射  

Facial Texture Mapping Based on Key Feature Points

作　　者：郑青碧[1] 毛自民[1] 

机构地区：[1]商丘职业技术学院计算机系,商丘476000

出　　处：《计算机与数字工程》2013年第1期111-114,共4页Computer & Digital Engineering

基　　金：河南省科技厅科技攻关项目(编号:112102210210);河南省教育厅自然科学研究计划项目(编号:2011A520034)资助

摘　　要：针对传统纹理映射方法中计算量大、鲁棒性差等问题,文章提出一种简便有效的纹理映射方法:首先在五官部位手工标定少量特征点,然后采用特征点标定算法和关键特征点三角剖分算法分别计算出全部顶点纹理坐标值和五官部位纹理坐标值,最后对五官纹理坐标值进行加权平滑运算并完成整个三维模型的纹理映射。实验结果表明,该方法不仅能有效消除映射误差,而且具有生成速度快、精度高、鲁棒性强的特点。Considering the high computational complexity, low robustness of traditional algorithm generating texture mapping, a simple and effective algorithm is proposed in this paper. Firstly, several feature points were selected by hand on facial features of 3D model and texture image. Then, the texture coordinates of all points and facial features were achieved according to feature points calibration algorithm and the triangulation algorithm of key feature points separately. Finally, the weighted smoothing algorithm was used to further calculate the texture coordinate of facial features, and the texture mapping of the entire 3D model was realized. The experiments show that this algorithm not only eliminates false mapping validly, but also maintains a good accuracy and robustness with a high speed.

关 键 词：三维人脸 特征点 纹理坐标 纹理映射 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]