3D reconstruction using projection of virtual height line based on dynamic programming

3D reconstruction is the base of mobile robot navigation. And it is necessary to provide the environment information for robots effectively and rapidly. The traditional 3D reconstruction method relies on the dense disparity map which requires high computing ability, and there is large redundant data in the reconstructed 3D data. Unlike traditional methods, the proposed method needn´t obtain the disparity map. We partitioned the scene scope uniformly into grids, and introduced virtual height lines(VHLs) which do not exist in the scene. Each VHL is projected in left image and right image. The intersection of VHL and the scene is only existent on projection line. So the point has maximum similarity metric. We used dynamic programming(DP) to find the matched lines in left image corresponding to the projections of VHLs(PVHLs) in right image. The intersection point of the matched line and the PVHL in left image is the answer. The experiments to reconstruct 3D scene are performed and the results show the proposed method is reliable and effective.