Joint Optimization of Spatial Registration and Histogram Compensation for Microscopic Images

An iterative registration algorithm, the Lucas-Kanade algorithm, is combined with the histogram transformation to jointly optimize the spatial registration and the histogram compensation. Based on a simple regression model, a nonparametric estimator, the empirical conditional mean, is used for the histogram transformation function. The proposed algorithm provides a good performance in registering microscopic images that have different exposure or histogram properties, and can easily adopt other histogram compensation schemes and variations of the Lucas-Kanade algorithms due to its implicit flexibility. Joint registration with a third-order polynomial warp and compensation is conducted for microscopic images that have different magnifications