3D Non-rigid Surface Matching and Registration Based on Holomorphic Differentials

摘要: 3D surface matching is fundamental for shape registration, deformable non-rigid tracking, recognition and classification. In this paper we describe a novel approach generating an efficient optimal combined from multiple boundary-constrained conformal parameterizations multiply connected domains (i.e., genus zero open with boundaries), which always come imperfect data acquisition (holes, partial occlusions, change of pose deformation between scans). This optimality criterion also used to assess how consistent each boundary is, thus decide enforce or relax constraints across the two surfaces be matched. The linear parameterization based on holomorphic differential forms, map n boundaries conformally planar rectangle (n - 2) horizontal slits, other as constraints. mapping diffeomorphism intrinsic geometry, handles arbitrary number boundaries, can implemented system. Experimental results are given real facial matching, cloth demonstrate efficiency our method, especially significantly inconsistent boundaries.