Skip to Main Content
Mutual information of images is widely used as similarity function in image registration. The image registration can be actually viewed as a multi-dimension space optimization problem, namely search the floating image transform parameters when mutual information is maximum. Yet the mutual information similarity function between two images is generally no convex and irregular, and parameters searching often results in local maximum. Considering this problem, particle swarm optimization algorithm is adopted in the medical image registration in this paper. Our experiment result shows that the improved particle swarm optimization based medical images registration can more easily achieve registration parameters of the global maximum with high robustness and precision.