A method for 3D-Point Reconstruction in multi-vision based on Perpendicular Foot

With increasing use of multi-vision system which has more than two cameras, 3D-Point Reconstruction in multi-vision become a basic issue in computer vision. This paper proposes a method for solving this problem. It takes the Perpendicular Foot of two Back-Projection Lines as the estimation of a 3D-Point, and then it sets the weights by analyzing error to find a better estimation. This method has a clear physical meaning and is relatively easy to be implemented in multi-vision systems, the simulation show, when the optical axes of cameras are roughly parallel, the accuracy and computational cost of this approach is better than traditional optimization algorithms.