Point Matching in the Presence of Outliers in Both Point Sets: A Concave Optimization Approach

Abstract

Recently, a concave optimization approach has been proposed to solve the robust point matching (RPM) problem. This method is globally optimal, but it requires that each model point has a counterpart in the data point set. Unfortunately, such a requirement may not be satisfied in certain applications when there are outliers in both point sets. To address this problem, we relax this condition and reduce the objective function of RPM to a function with few nonlinear terms by eliminating the transformation variables. The resulting function, however, is no longer quadratic. We prove that it is still concave over the feasible region of point correspondence. The branch-and-bound (BnB) algorithm can then be used for optimization. To further improve the efficiency of the BnB algorithm whose bottleneck lies in the costly computation of the lower bound, we propose a new lower bounding scheme which has a k-cardinality linear assignment formulation and can be efficiently solved. Experimental results show that the proposed algorithm outperforms state-of-the-arts in its robustness to disturbances and point matching accuracy.

An Efficient Multilinear Optimization Framework for Hypergraph Matching

“…For future work it might be interesting to go beyond injective mappings which require every point on one image to be mapped to exactly one point in the second image. Non-injective mappings are of interest if not all points need to be matched as for instance when outliers are present on both sides [30], [40]. Our Theorem 3.3 allows the use of new constraint sets reflecting non-injective mappings, which is an interesting topic of further research.…”

Feature Guided Biased Gaussian Mixture Model for image matching

“…Lian and Zhang [35] proposed a concave optimization approach for the robust point matching problem when no outliers exist. They further extend their work to handle outliers in [36] by reducing the original object function to a function with fewer nonlinear terms. In spite of the success in point sets registration, the methods mentioned above are seldom used in image matching.…”

Convex hull indexed Gaussian mixture model (CH-GMM) for 3D point set registration

“…Chui et al [37][38][39] proposed TPS robust point matching to estimate non-rigid transformations between point sets by using spatial mapping and outlier rejection to establish correspondences and to optimize warping parameters. Lian et al [40,41] suggested that the traditional robust point matching method could be reduced to a concave function by eliminating transformation variables and applying linear transformation. Concave optimization benefits correspondence problems and achieves a globally optimal solution without regularizing transformations.…”