A Neighborhood Incorporated Method in Image Registration

Mutual information has been widely used in image registration as an effective similarity measure. It has attracted a lot of attention to the effective use of the spatial information. Here we propose a new measure that includes the mean of the neighborhood region of each pixel as one variable of the two-dimension normal distribution assumed in our method. The experimental results show that our method can not only improve the robustness of mutual information, but also reduce the affect of noise in image registration.