3D Object Classification Using Scale Invariant Heat Kernels with Collaborative Classification

One of the major goals of computer vision is the development of flexible and efficient methods for shape representation. This paper proposes an approach for shape matching and retrieval based on scale-invariant heat kernel (HK). The approach uses a novel descriptor based on the histograms of the scale-invariant HK for a number of critical points on the shape at different time scales. We propose an improved method to introduce scale-invariance of HK to avoid noise-sensitive operations in the original method. A collaborative classification (CC) scheme is then employed for object classification. For comparison we compare our approach to well-known approaches on a standard benchmark dataset: the SHREC 2011. The results have indeed confirmed the high performance of the proposed approach on the shape retrieval problem.