3D reconstruction of indoor and outdoor scenes using a mobile range scanner

摘要

Describes 3D mapping of indoor and outdoor environments using a mobile range scanner. In the raw range data preprocessing stage, we propose to use area decreasing flow for surface mesh smoothing instead of the mean curvature flow. Although the proposed area decreasing flow approach is mathematically equivalent to the mean curvature flow, it can avoid the difficulty in curvature estimation and provide an optimal flowing step size. Geometric details are preserved by adaptive smoothing on crease edges. In the multi-view 3D reconstruction stage, we combine the space carving with Hilton's implicit surface-based method (Hilton et al., 1998) to generate watertight 3D models from limited number of scans. Volumetric deformation by mean curvature flow using the level set method is applied in the post-processing stage to remove outliers. We present results of 3D reconstruction of real indoor and outdoor environments from multi-view noisy range data.