. . . "where A=1+D, where D is a deformation matrix and d is the translation of the feature window. Such an approach is described in computer vision literature such as Lucas B. D. and Kanade, T. "An Iterative Image Registration Technique and" .