Bag-of-feature-graphs: A new paradigm for non-rigid shape retrieval

This paper advocates a new paradigm, called bag-of-feature-graphs (BoFG), for non-rigid shape retrieval. It represents a shape by constructing graphs among its features, which significantly reduces the number of points involved in computation. Given a vocabulary of geometric words, for each word the BoFG builds a graph that records spatial information of features, weighted by their similarities to this word. This eliminates unlikely points in a word category, during shape comparison. Feature graphs are governed by their affinity matrices of weighted heat kernels, whose eigenvalues form a concise shape descriptor. Evaluations of the proposed method are conducted via quantitative measurements. The results demonstrate that the BoFG has competitive precisions w.r.t. state-of-the-art methods, and is much faster to compute.