An automatic 3D reconstruction method based upon coplanar and region constraints is proposed for synthesizing piecewise 3D planar models of a real-world scene portrayed in a sequence of digital images. Significantly, the reconstruction process requires no prior knowledge about the scene and the camera parameters. The proposed modeling method comprises three steps. Firstly, the corresponding feature points and lines are extracted from a series of images of the world scene. Secondly, the extracted corresponding points and lines are filtered in accordance with region and coplanar constraints and are used to identify the possible half-planes of real-world planes. Finally, a complete 3D planar model is constructed by identifying the correct half-planes in the world scene enlarging these half-planes to their full extent, and then merging all the extended halfplanes which from part of the same world plane. The feasibility of the proposed approach is demonstrated by reconstructing 3D planar models of three real-world scenes containing objects with multiple planar facets.
|頁（從 - 到）||361-378|
|期刊||Journal of Information Science and Engineering|
|出版狀態||Published - 2013 3月|
All Science Journal Classification (ASJC) codes