Mutual Information for Lucas-Kanade Tracking (MILK): An Inverse Compositional Formulation

Mutual information (Ml) is popular for registration via function optimization. This work proposes an inverse compositional formulation of Ml for Levenberg-Marquardt optimization. This yields a constant Hessian, which may be precomputed. Speed improvements of 15 percent were obtained, with convergence accuracies similar those of the standard formulation.