Image registration by automatic subimage selection and maximization of combined mutual information and spatial information

Image registration is one of the most important steps in the analysis of remotely sensed data. Mutual information is a robust similarity metric in image registration. Unfortunately mutual information neglects spatial information. In this work, we propose a new similarity metric for image registration called enhanced mutual information (EMI), which combines mutual information with a weighting function based on the absolute difference of corresponding pixel values. We also use subimages with high entropy as a search data strategy Experimental results show that our proposed method was more robust to noise and accurate than the standard methods used.