CONCAVE OPTIMIZATION APPROACH FOR MINIMIZATION OF ENERGY FUNCTION USING RPM
Yamala Jagadeesh, B.Jayanth Nadh · IJCER · 2013
An essential yet demanding problem in computer visualization, pattern identification and medical image examination is a point matching. The well-known robust point matching (RPM) method uses deterministic strengthen for optimization, and it has two troubles. First, it cannot assure the worldwide optimality of the solution and tends to support the centres of two point sets. Second, in order to avoid the generation of undesirable results deformation needs to be normalized. Here, after eliminating the transformation variables and applying linear transformation, we first show that the energy function of RPM can be reduced to a concave function with very few non-rigid terms to address the above problem; and to minimize the resulting energy function we then propose to use concave optimization technique. For simple transformations such as similarity transform, the proposed method scales well with difficulty size, accomplishes the worldwide best solution, and does not need regularization. Experiments on artificial and real data authenticate the benefits of our method in contrast with state-of-the-art methods.