基于直线光流场的三维运动和结构重建

利用直线间运动对应关系，将像素点光流的概念和定义方法应用于直线，提出了直线光流的概念，建立了求解空间物体运动参数的线性方程组，利用三幅图像21条直线的光流场，可以求得物体运动的12个参数以及空间直线坐标．但是在实际应用当中，要找出这21条直线的光流场是很困难的，因此该文提出了运用解非线性方程组的方法，只需要6条直线的光流．就可以分步求出物体的12个运动参数，并根据求得的12个运动参数和一致的图像坐标系中的直线坐标，求得空间直线的坐标，从而实现了三维场景的重建．

Reconstructing 3D Motion and Structure from Straight-Line Optical Flow

A mathematical definition of straight-line optical flow is first presented in this paper, and according to the conception and definition of pixel optical flow, we define the time derivative of image line coordinates as straight-line optical flow. Then linear algorithm for reconstructing 3D motion and structure from straight-line optical flow is presented. The method is to use optical flow of 21 straight lines in 3 consecutive image frames to establish a set of equations about 3D motion parameters. After solving the above equations, 12 motion parameters of 3D object and coordinates of the 3D straight line could be obtained. In fact, it is difficult to get the 21 straight lines and their correspondence. Another non-linear algorithm is presented. The method is to use optical flow of 6 straight-line in 3 consecutive image frames to solve the problem of structure from motion. When applying the method of solving non-linear equations, only 6 equations are needed to get rotation and translation parameters. Then the 3D coordinates of the straight line can be recovered.