用单数码相机实现物体表面的三维重建  

3D object surface reconstruction using single camera

作　　者：张勇[1] 金学波[1] 

机构地区：[1]浙江理工大学信息电子学院,浙江杭州310018

出　　处：《计算机工程与设计》2008年第11期2926-2929,共4页Computer Engineering and Design

基　　金：国家自然科学基金项目(60674028);浙江省教育厅重点专业基金项目(111231A3255401)

摘　　要：根据由运动重建物体结构的原理,设计了一个简便易操作的三维重建系统,具体做法是:先用张氏标定法求得内参数矩阵,然后在两个不同的未知位置拍摄物体得到两幅图像,经立体匹配后,利用图像特征点的对应关系求解基本矩阵和本质矩阵,分解本质矩阵获得两个拍摄位置确定的摄像机运动参数(旋转矩阵和平移向量),进而求出相机在两个位置的投影矩阵,最后用三角法计算出物体表面特征点的三维坐标并在OpenGL中重建物体表面。和传统的立体视觉系统相比,本系统只需要一台数码相机和平面方格模板就可以实现三维重建,因此适用于普通相机用户。A flexible 3D reconstruction technique to easily used is proposed, based on structure from motion. The concrete method is shown as follows： the camera is calibrated and the intrinsic parameters is sotten using Zhang calibration method, then, the object in two different orientations and the motion need not to be known, after matching the two images robustly, the fundamental matrix and es- sential matrix from point correspondences are estimated, further, by decomposing essential matrix, the motion parameters （R, t） de- termined by the two different orientations is solved and the projective matrixes corresponding the two orientations is obtained. In the end, the 3D coordinates of the object feature points is computed and the surface of the object in OpenGL is displaied. Compared with the clas- sical binocular vision system which use expensive equipments, this technique only requires a digital camera and a planar pattern, it is well suited for general digital camera users without specialized knowledge of 3D geometry.

关 键 词：相机标定 三维重建 立体匹配 本质矩阵 三角剖分 纹理粘贴 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]