Automatic weights determination by NLSSQP method in phase unwrapping

This paper presents a possible automatic approach to back analyze the weights by incorporating phase unwrapping program into a minimization method NLSSQP. The proposed approach is illustrated through two simulative examples corrupted by uniformly distributed noise and scattered noise. By introduction of real-valued weights, which are computed by NLSSQP method, unwrapped phases are converged to correct solution. Real-valued weights obtained in this approach turned out to be superior to customary binary-valued weights when images are corrupted by noise at the edges or corners