This paper introduces a sparse and incremental 2-manifold surface reconstruction method. It uses a sparse 3D point cloud generated by a Structure-from-Motion algorithm (SfM) as its main input as opposed to the more common dense algorithms. Furthermore, our method is incremental: the surface is updated for every new camera pose computed by SfM, and the update occurs in a small neighborhood of the new camera pose. Compared to the other surface reconstruction methods, our method has the advantage to have all these properties at the same time. The quality and execution time of the proposed algorithm is evaluated on a large scale (2.5 km.) real sequence taken in an urban environment, and the method is quantitatively evaluated on a synthetic urban scene.

Abstract-In the recent years, a family of 2-manifold surface reconstruction methods from a sparse Structure-from-Motion points cloud based on 3D Delaunay triangulation was developed. This family consists of batch and incremental variations which include a step that remove visual artifacts. Although been necessary in the term of surface quality, this step is slow compared to the other parts of the algorithm and is not well suited to be used in an incremental manner. In this paper, we present two other methods for removing visual artifacts. They are evaluated and compared to the previous one in the incremental context where the need of new methods is the highest. Taken separately, they provide medium results, but used together they are as good as the old method in the terms of surface quality, and at the same time, processing time is almost three times smaller.

No abstract

The majority of methods for the automatic surface reconstruction of a scene from an image sequence have two steps: Structure-from-Motion and dense stereo. From the complexity viewpoint, it would be interesting to avoid dense stereo and to generate a surface directly from the sparse features reconstructed by SfM. This paper adds two contributions to our previous work on 2-manifold surface reconstruction from a sparse SfM point cloud: we quantitatively evaluate our results on standard multiview dataset and we integrate the reconstruction of image curves in the process.