3D Reconstruction based on hierarchical level projection of virtual height line

3D reconstruction is the essential part of robots' autonomous navigation. 3D reconstruction method based on hierarchical level projection of virtual height line (PVHL) is proposed. Unlike traditional method, the proposed method could reconstruct 3D scene independent with the dense disparity map. At first we partitioned the interesting scope uniformly into grid with a certain size, and introduced virtual height lines (VHLs) which do not exist in the scene. Each VHL is projected in left and right image. The intersection of VHL and the scene is only existent on projection lines. So the point has maximum similarity metric, and the height of the point is that of the corresponding scene. Finding the height of the scene is converted to prove whether the point of intersection has the maximum similarity metric or not. Then we reconstructed the scene in the grids with an initial size and generated a coarse reconstructed scene. By partitioning and reconstructing the grids hierarchically, we could refine the coarse result, and obtain more accurate information of the scene. The experiment to reconstruct 3D scene are performed and the results show the proposed method is effective.