In this paper, we consider the problem of finding the feature correspondencesamong a collection of feature sets, by using their point-wise unary features.This is a fundamental problem in computer vision and pattern recognition, whichalso closely relates to other areas such as operational research. Differentfrom two-set matching which can be transformed to a quadratic assignmentprogramming task that is known NP-hard, inclusion of merely unary attributesleads to a linear assignment problem for matching two feature sets. Thisproblem has been well studied and there are effective polynomial global optimumsolvers such as the Hungarian method. However, it becomes ill-posed when theunary attributes are (heavily) corrupted. The global optimal correspondenceconcerning the best score defined by the attribute affinity/cost between thetwo sets can be distinct to the ground truth correspondence since the scorefunction is biased by noises. To combat this issue, we devise a method formatching a collection of feature sets by synergetically exploring theinformation across the sets. In general, our method can be perceived from a(constrained) clustering perspective: in each iteration, it assigns thefeatures of one set to the clusters formed by the rest of feature sets, andupdates the cluster centers in turn. Results on both synthetic data and realimages suggest the efficacy of our method against state-of-the-arts.
展开▼
