Abstract:
We present a novel convex programming scheme to solve matching problems, focusing on the challenging problem of matching in a large search range and with cluttered background. Matching is formulated as metric labeling with L1 regularization terms, for which we propose a novel linear programming relaxation method and an efficient successive convexification implementation. The unique feature of the proposed relaxation scheme is that a much smaller set of basis labels is used to represent the original label space. This greatly reduces the size of searching space. A successive convexification scheme solves the labeling problem in a coarse to fine manner. Importantly, the original cost function is re-convexified at each stage, in the new focus region only, and the focus region is updated so as to refine the searching result. This makes the method well suited for large label set matching. Experiments demonstrate successful applications of the proposed matching scheme in object detection, motion estimation and tracking.
Citations
|
4776
|
Probabilistic reasoning in intelligent systems: networks of plausible inference
– Pearl
- 1988
|
|
1268
|
Equations of state calculations by fast computing machines
– Metropolis, Rosenbluth, et al.
- 1953
|
|
600
|
On the statistical analysis of dirty pictures
– Besag
- 1986
|
|
527
|
Visual Reconstruction
– Blake, Zisserman
- 1987
|
|
517
|
Fast approximate energy minimization via graph cuts
– Boykov, Veksler, et al.
- 2001
|
|
249
|
Scene labeling by relaxation operations
– Rosenfeld, Hummel, et al.
- 1976
|
|
179
|
A graduated assignment algorithm for graph matching
– Gold, Rangarajan
- 1996
|
|
176
|
A maximum-flow formulation of the n-camera stereo correspondence problem
– Roy, Cox
- 1998
|
|
150
|
Multi-camera scene reconstruction via graph cuts
– Kolmogorov, Zabih
|
|
146
|
Computing visual correspondence with occlusions using graph cuts
– Kolmogorov, Zabih
|
|
121
|
Stereo matching using belief propagation
– Sun, Zheng, et al.
- 2003
|
|
98
|
Efficient belief propagation for early vision
– Felzenszwalb, Huttenlocher
|
|
95
|
A new algorithm for non-rigid point matching
– Chui, Rangarajan
- 2000
|
|
93
|
Approximation algorithms for classification problems with pairwise relationships: metric labelingand markov random fields
– Kleinberg, Tardos
- 1999
|
|
83
|
On the optimality of solutions of the max-product belief propagation algorithm in arbitrary graphs
– Freeman, Weiss
- 2001
|
|
69
|
Computing Geodesics and Minimal Surfaces via Graph Cuts
– Boykov, Kolmogorov
|
|
57
|
Finding deformable shapes using loopy belief propagation
– Coughlan, Ferreira
- 2002
|
|
39
|
Approximation algorithms for the metric labeling problem via a new linear programming formulation
– Chekuri, Khanna, et al.
|
|
24
|
Representation and detection of deformable shapes
– Felzenszwalb
|
|
20
|
Global optimization using embedded graphs
– Ishikawa
- 2000
|
|
15
|
RealTime Motion Analysis with Linear Programming
– Ben-Ezra, Peleg, et al.
- 1999
|
|
10
|
Numerical solutions by the continuation method
– Wasserstrom
- 1973
|
|
6
|
Optimizing motion estimation with linear programming and detail-preserving variational method
– Jiang, Li, et al.
- 2004
|
|
5
|
A comparison of search strategies for geometric branch and bound algorithms
– Breuel
- 2002
|
|
3
|
Linear programming matching and appearance-adaptive object tracking”, Energy Minimization Methods
– Jiang, Drew, et al.
- 2005
|
|
2
|
Stability of Lagrangian duality for non convex quadratic programming solution methods and applications to computer vision”, Mathematical Modelling and Numerical Analysis
– Tao, Phong, et al.
- 1997
|
|
1
|
Graph matching using embedding and semidefinite
– Bai, Yu, et al.
- 2004
|
