Robust Point Matching Revisited: A Concave Optimization Approach



We proposed a new approach to minimizing the energy function of the classical RPM method. After eliminating the transformation variable, we reduced the energy function to a concave quadratic program, which can be efficiently solved by large scale concave optimization techniques.

Our method can guarantee the global optimality of the solution, and does not need to regularize deformation for simple transformations such as similarity transform. Our method also scales well with problem size due to the special structure of its optimization problem

Extensive experimental results demonstrated that the proposed method has high matching accuracy and high robustness to disturbances, such as clutter, in comparison with state-of-the-art methods.

Regularized Robust Coding for Face Recogntion




There are no reviews yet.

Be the first to review “Robust Point Matching Revisited: A Concave Optimization Approach”

Your email address will not be published. Required fields are marked *