An Adaptive Descriptor for Functional Correspondence Between Non-Rigid 3D Shapes

摘要

In this paper, we propose an adaptive descriptor for functional correspondence between non-rigid 3D shapes. This descriptor is computed based on a norm between points in Global Point Signature (GPS) domain. We calculate the initial descriptors by utilizing initial correspondences estimating in GPS domain. Moreover, we present an iteration method for extending the initial descriptor based on candidate correspondences. Our method is implemented and benchmarked with others; the results shows that our paradigm improves overall matching canability.