Registration by automatic subimage selection and maximization of mutual information

Image registration is one of the crucial steps in the processing of remotely sensed data. It is the overlaying of two images of the same scene taken at different times or by different sensors. The search for the matching transformation can be very tedious and time consuming. For high accuracy and robustness as well as low computational cost, a suitable similarity metric and reduction in search data and search space is needed. In this paper, we investigate mutual information as similarity metric. The sensitivity of mutual information and correlation coefficient to noise is demonstrated. Furthermore, we show the effect of bin size on mutual information. We also look into subimage selection as reduction in search data strategy. We propose a measure, called alignability, which shows the ability of die subimage to provide robust registration results. Alternative subimage selection methods, such as entropy and gradient magnitude, are also presented. Alignability is compared with information contents and edge contents respectively.