摘要
提出了一种从两幅未标定图象重建场景三维模型的方法 .这种方法充分利用了人造结构场景中大量存在的平行性和正交性几何约束 ,即利用每幅视图中三组互相垂直的平行线 ,计算出 3个影灭点 ,从而对每幅视图进行标定 .对两幅未标定图象 ,从基本矩阵只能得到射影重构 ,如果每幅图象都已标定 ,则可将基本矩阵转化为本质矩阵 .三维重构过程有两个步骤 :先是恢复相机的位置和运动 ;后是用三角测量法计算出点的三维坐标 .对多平面组成的场景进行三维重构实验 ,所得三维模型产生新的视点图象 ,与所观察的场景一致 ,重构的两个平面夹角与实际值相近 ,实验结果表明 。
In this paper, the methods for the uncalibrated two-views 3D reconstruction is proposed. The methods employ geometric constraints available from geometric relationships that are plentiful in manmade structure-such as parallelism and orthogonality of lines and planes, these constraints lead to simple method to calibrate the intrinsic parameters of the camera. This is done by determining the vanishing points associated with parallel lines in the world, under the assumption of zero skew and known aspect ratio, three mutually orthogonal directions are exploited to give the camera calibration matrix . It is possible to obtain only a projective reconstruction from the fundamental matrix. If each image is calibrated, it be able to convert from the uncalibrated fundamental matrix to the essential matrix. A Euclidean reconstruction would be preferable. Once the essential matrix is recovered, if the first camera is assumed to be at origin of the coordinate system, then it is a simple matter to calculate the rotation and translation of the second camera relative to the first. After the camera intrinsic and extrinsic parameters had been estimated, camera projection matrices may be recovered and used to estimate the structure. The 3D reconstruction process has two stages: the first to recover the camera positions and motions, the second step involves triangulation to recover the 3D points. The validity of the proposed algorithm is confirmed by experiment for a number of multi-plannar scenes. The reconstructed scene is modeled. New images are generated of the model for new view points. The geometry agrees with our perception of scene. The angle between the two reconstructed planes looks just like a right angle.
出处
《中国图象图形学报（A辑）》
CSCD
北大核心
2003年第8期872-876,共5页
Journal of Image and Graphics
基金
国防科工委预研基金 (B0 12 2 -0 3 5 )
关键词
三角测量法
图象标定
几何约束
影灭点
本质矩阵
基本矩阵
计算机视觉
Computer image processing,Vanishing points, Camera calibration, Fundamental matrix, Essential matrix, Triangulation, 3D reconstruction
