Dense multi-planar scene reconstruction from sparse point cloud

There are multi-planar scenes everywhere in our daily life. However, given its lack and self-repeat of the texture, there would be problems of over scarcity and holes on the reconstructed point cloud by the method of multi-view reconstruction. Further, there would be vacillation over the reconstructed facades using the method of fitting the reconstructed point cloud with miniature facets. To address these problems, we propose a method of piecewise reconstruction of each plane from the sparse point cloud. The proposed method first improves the J-linkage algorithm, with the stratified sampling instead of the random sampling. We then fit the point cloud with planes using the improved J-linkage algorithm, to obtain the multi-planar model of the scene. Finally, we extract and reconstruct the planar regions with the multi-planar model as well as an unsupervised segmentation algorithm. Besides, the non-planar areas are reconstructed by using the clustering views for multi-view stereo/patch-based multi-view stereo (CMVS/PMVS) algorithm. Experimental results of the multi-planar model demonstrate that the improved J-linkage algorithm can enhance the accuracy of the multi-planar model. Also, the experimental results of 3D reconstruction show that our method not only can effectively overcome holes and jaggies problems, but also can model the complete planar regions. Copyright © 2015 Acta Automatica Sinica. All rights reserved.