Generalized Local-to-Global Shape Feature Detection Based on Graph Wavelets

摘要: Informative and discriminative feature descriptors are vital in qualitative quantitative shape analysis for a large variety of graphics applications. Conventional primarily concentrate on discontinuity certain differential attributes at different orders that naturally give rise to their power depicting point, line, small patch features, etc. This paper seeks novel strategies define generalized, user-specified features anywhere shapes. Our new region-based constructed with the powerful spectral graph wavelets (SGWs) both multi-scale multi-level nature, incorporating local (differential) global (integral) information. To our best knowledge, this is first attempt organize SGWs hierarchical way unite them bi-harmonic diffusion field towards analysis. Furthermore, we develop local-to-global detection framework facilitate host applications, including partial matching without point-wise correspondence, coarse-to-fine recognition, model Through extensive experiments comprehensive comparisons state-of-the-art, has exhibited many attractive advantages such as being geometry-aware, robust, discriminative, isometry-invariant,