一种基于奇异值分解的分层重构算法  

An Algorithm of Hierarchical Reconstruction Based on SVD

作　　者：梁栋[1] 韦穗[1] 吴福朝[1] 杨尚俊[2] 

机构地区：[1]安徽大学计算智能与信号处理教育部重点实验室 [2]安徽大学数学系,安徽合肥230039

出　　处：《中国科学技术大学学报》2001年第5期571-576,共6页JUSTC

基　　金：国家 8 6 3计划资助项目 (86 3 980 4 0 9)

摘　　要：以仿射投影来逼近透视投影 ,采用共轭梯度法迭代估计射影深度 ,通过测量矩阵的奇异值分解实现射影重构 .在摄象机内参数已知的情况下 ,求解一个满足欧氏重构条件的 4× 4非奇异矩阵 ,由此矩阵将射影重构变换为欧氏重构 .Affine projection is used to approximate perspective projection, projective depths are iteratively estimated by using the conjugate gradient method, and the projective reconstruction is obtained by SVD of the measurement matrix. Under the circumstances in which the intrinsic parameters of camera are known, a 4x4 non singular matrix satisfying the conditions of euclidean reconstruction is evaluated, with which projective reconstruction is transformed into euclidean reconstruction. The result indicates that the algorithm is effective.

关 键 词：奇异值分解 3D重构 计算机视觉 分层重构算法 射影重构 射影深度 矩阵分解 

分 类 号：TN911.73[电子电信—通信与信息系统]