Particle Swarm Optimization with Powell´s Direction Set Method for Remote Sensing Image Registration

Multisensor image registration via mutual information is a powerful approach. However, there are a lot of local maxima in the mutual information surface which prevent optimization methods from finding the global maximum. To solve the problem, the particle swarm optimization (PSO, global optimizer) is used to find a good initial estimation for registration parameters and followed by a refining Powell´s direction set method (PDSM, local optimizer). Experiments with simulated and real multisensor images demonstrate the effectiveness of the proposed optimization method with PSO and PDSM for multisensor image registration.