【作者】 梁栋； 阮宗才； 韦穗；

【Author】 Liang Dong Ruan Zongcai Wei Sui Educational Department Key Lab. of IC&SP, Anhui University, Hefei 230039

【机构】 安徽大学计算智能与信号处理教育部重点实验室；

【摘要】 本文提出一种多视点3D重构算法。首先,利用基础矩阵和极点计算射影深度;第二,构建测量矩阵并通过奇异值分解将测量矩阵分解为射影空间下的摄像机运动和物体形状(射影重构);第三,在摄像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,通过此矩阵可以将射影空间下的摄像机运动和物体形状变换为欧氏空间下的摄像机运动和物体形状(欧氏重构)。更多还原

【Abstract】 In this paper, an algorithm is proposed for 3D reconstruction from point correspondences observed in multiple images. First, the projective depths are computed by using fundamental matrices and epipoles obtained from image data. Second, the measurement matrix is constructed and the projective reconstruction is obtained by decomposing the measurement matrix in SVD. Third, in the case of the intrinsic parameters of camera being known, a 4×4 non-singular matrix is evaluated, which satisfies the conditions of euclidean reconstruction. This matrix can upgrade projective reconstruction into euclidean reconstruction.更多还原