Rigid Surface Matching by Shape Analysis and Correspondences Filtering

摘要

Surface matching technology, as a fundamental part of computer graphics, has a wide range of applications in engineering domain. In this paper, we propose a novel matching method for best fitting a given source model onto a target model by a combination of the following: point clustering, region similarity matching and correct pairing searching. Mesh segmentation is used to aggregate points into several sub-regions, and then paired regions are obtained by distance distribution. An indirect searching strategy is presented, in which bad paired triples are removed by filtering operation and the remaining candidates are traversed until the correct pairing is found. Experimental results on a variety of mechanical models demonstrate that our method exhibits high reliability and robustness for automatic shape alignment.