摘要
提出一种改进的三维人脸重构方法.该方法采用基于单个相机的双目立体视觉系统对人脸进行采样,根据人脸对称性假设,运用补洞与纠错技术进行自动点云优化.继而采用简化的Candide-3模型作为细分初始控制网格,局部加细地进行细分曲面分层次拟合操作,采用测地线映射技术对不同表情进行归一化,并分别建立人脸数据库.实验结果表明,采用单相机立体视觉系统在提高重建精度的同时,很大程度上避免由于双相机拍摄不同步引起的重建鲁棒性降低问题.而采用细分曲面作为存储结构,在节约空间的前提下,为分层次比对筛选提供理论支持.该系统成本较低,适合在许多领域推广应用.
An improved 3D face reconstruction method as well as a binocular stereo vision system based on single camera is proposed. Under the assumption that face is symmetrical, the point cloud is optimized automatically by correction and holes filling. Then, a simplified Candide-3 model is used as initial subdivision controlling mesh, locally refined and levelly fitted. Meanwhile, geodesic mapping technique is applied to normalize different expressions and face databases are built respectively. Experimental results show that the proposed stereo vision system improves the reconstruction accuracy and avoids robust decreasing caused by non synchronous shooting of two cameras. Furthermore, subdivision surfaces used as storage saves space and provides theoretical support for comparison. Considering its low cost, the proposed system is feasible to spread in many fields.
出处
《模式识别与人工智能》
EI
CSCD
北大核心
2010年第5期686-694,共9页
Pattern Recognition and Artificial Intelligence
基金
浙江省重大科技专项重点工业项目(No.2009C11023)
浙江省科技计划项目(No.2009C31120
2009C34006
2008C21084)
浙江省自然科学基金项目(No.Y1100440)
浙江省教育厅项目(No.Y200804427
Y200907886)资助
关键词
双目视觉
三维人脸
重构
细分法
Binocular Stereo Vision, 3D Face, Reconstruction, Subdivision Method
