Joint Optimization of Image Registration and Comparametric Exposure Compensation Based on the Lucas-Kanade Algorithm

Abstract

An iterative registration algorithm, the Lucas-Kanade algorithm, is combined with an exposure compensation algorithm to jointly optimize the spatial registration and the exposure compensation. The coordinate descent method is employed to minimize a mean squared error between image pairs. Based on a simple regression model, a non-parametric estimator, the empirical conditional mean and its polynomial fitting are used as histogram transformation functions for the exposure compensation. The proposed algorithm performs a good registration for real perspective and microscopic images, and can easily adopt other exposure compensation approaches and variations of the Lucas-Kanade algorithms due to its implicit flexibility