3D reconstruction based on single-lens stereovision system using biprism

The objective of this paper is to reconstruct three-dimensional scenes from stereo images. In this paper, the single-lens stereovision system using biprism is employed to capture image pairs. Then, we propose a geometry-based approach to solve stereo rectification problem based on this system. It does not require complex calibration at all and also can obtain accurate extrinsic matrix. Then we propose an algorithm of stereo matching to obtain the disparity map. This method can filter outliers and deal with occlusion. Lastly, we use the algorithm of triangulation to calculate the 3D point cloud from the disparity map and then the final 3D scene is reconstructed. The experimental results show that our proposal methods are feasible and effective.