A Structural Matching Algorithm Using Generalized Deterministic Annealing
We present a practical implementation of a structural matching algorithm that uses the generalized deterministic annealing theory. The problem is formulated as follows: given a set of model points and object points, find a matching algorithm that brings the two sets of points into correspondence. An “energy” term represents the distance between the two sets of points. This energy has many local minima and the purpose is to escape from these local minima and to find the global minimum using the simulated annealing theory. We use a windowed implementation and a suitable definition of the energy function that reduces the computational effort of this annealing schedule without decreasing the solution quality.
Keywordsgeneralized deterministic annealing structural matching unary and binary features graph matching local Markov chain
- 1.S. Acton. A. Bovik, “Generalized deterministic annealing”. IEEE Transactions on Neural Networks May 1996Google Scholar
- 2.S. Gold, “A graduate assignment algorithm for graph matching”, IEEE Transactions on Pattern Anal. And Machine Intell. April 1996Google Scholar
- 3.A. Rangarajan, “A Lagrangian Relaxation Network for Graph Matching, IEEE Transactions on Neural Networks, Nov 1996Google Scholar