A quadratic penalty method for hypergraph matching

Hypergraph matching is a fundamental problem in computer vision. Mathematically, it maximizes a polynomial objective function, subject to assignment constraints. In this paper, we reformulate the hypergraph matching problem as a sparse constrained optimization problem. By dropping the sparse constraint, we show that the resulting relaxation problem can recover the global minimizer of the original problem. This property heavily depends on the special structures of hypergraph matching. The critical step in solving the original problem is to identify the location of nonzero entries (referred to as the support set) in a global minimizer. Inspired by such observation, we apply the quadratic penalty method to solve the relaxation problem. Under reasonable assumptions, we show that the support set of the global minimizer in a hypergraph matching problem can be correctly identified when the number of iterations is sufficiently large. A projected gradient method is applied as a subsolver to solve the quadratic penalty subproblem. Numerical results demonstrate that the exact recovery of the support set indeed happens, and the proposed algorithm is efficient in terms of both accuracy and CPU time.

[1]  D. Dai,et al.  PRL‐dock: Protein‐ligand docking based on hydrogen bond matching and probabilistic relaxation labeling , 2012, Proteins.

[2]  Naihua Xiu,et al.  A convergent iterative hard thresholding for nonnegative sparsity optimization , 2017 .

[3]  Yosi Keller,et al.  A Probabilistic Approach to Spectral Graph Matching , 2013, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[4]  Christian Kanzow,et al.  Mathematical Programs with Cardinality Constraints: Reformulation by Complementarity-Type Conditions and a Regularization Method , 2016, SIAM J. Optim..

[5]  Yonina C. Eldar,et al.  Sparsity Constrained Nonlinear Optimization: Optimality Conditions and Algorithms , 2012, SIAM J. Optim..

[6]  Yong Zhang,et al.  Sparse Approximation via Penalty Decomposition Methods , 2012, SIAM J. Optim..

[7]  João Paulo Costeira,et al.  A Global Solution to Sparse Correspondence Problems , 2003, IEEE Trans. Pattern Anal. Mach. Intell..

[8]  Wei Liu,et al.  Discrete hyper-graph matching , 2015, 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[9]  Paul H. Calamai,et al.  Projected gradient methods for linearly constrained problems , 1987, Math. Program..

[10]  Ya-Xiang Yuan,et al.  Optimization Theory and Methods: Nonlinear Programming , 2010 .

[11]  Minsu Cho,et al.  Hyper-graph matching via reweighted random walks , 2011, CVPR 2011.

[12]  Naihua Xiu,et al.  Optimality conditions for sparse nonlinear programming , 2017 .

[13]  Hong Yan.  Efficient Matching and Retrieval of Gene Expression Time Series Data Based on Spectral Information , 2005, ICCSA.

[14]  Naihua Xiu,et al.  On Solutions of Sparsity Constrained Optimization , 2015 .

[15]  Michael S. Brown,et al.  As-Projective-As-Possible Image Stitching with Moving DLT , 2013, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[16]  Jong-Ha Lee,et al.  Topology Preserving Relaxation Labeling for Nonrigid Point Matching , 2011, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[17]  Jean Ponce,et al.  A tensor-based algorithm for high-order graph matching , 2009, 2009 IEEE Conference on Computer Vision and Pattern Recognition.

[18]  D. Bertsekas.  Projected Newton methods for optimization problems with simple constraints , 1981, CDC 1981.

[19]  Amnon Shashua,et al.  Probabilistic graph and hypergraph matching , 2008, 2008 IEEE Conference on Computer Vision and Pattern Recognition.

[20]  Jitendra Malik,et al.  Shape matching and object recognition using low distortion correspondences , 2005, 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05).

[21]  Xue Li,et al.  The First-Order Necessary Conditions for Sparsity Constrained Optimization , 2015 .

[22]  Quynh N. Nguyen,et al.  An Efficient Multilinear Optimization Framework for Hypergraph Matching , 2015, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[23]  Roger Fletcher,et al.  New algorithms for singly linearly constrained quadratic programs subject to lower and upper bounds , 2006, Math. Program..

[24]  Heinz H. Bauschke,et al.  Restricted Normal Cones and Sparsity Optimization with Affine Constraints , 2012, Found. Comput. Math..

[25]  Ya-Feng Liu,et al.  Lp-norm Regularization Algorithms for Optimization Over Permutation Matrices , 2016, SIAM J. Optim..

[26]  Hong Yan,et al.  Local Topology Preserved Tensor Models for Graph Matching , 2015, 2015 IEEE International Conference on Systems, Man, and Cybernetics.

[27]  Lei Guo,et al.  An Augmented Lagrangian Method for Non-Lipschitz Nonconvex Programming , 2017, SIAM J. Numer. Anal..

[28]  Ze-Nian Li,et al.  Matching by Linear Programming and Successive Convexification , 2007, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[29]  Stephen J. Wright,et al.  Numerical Optimization , 2018, Fundamental Statistical Inference.

[30]  Christian Kanzow,et al.  Constraint qualifications and optimality conditions for optimization problems with cardinality constraints , 2016, Math. Program..