Optimal Homography Computation with a Reliability Measure

normalization to make the Z component 1. In order This paper describes an algorithm for optimally cornto remove scale we the norputing the homography between two given sets of points 1' = ', where the norm of matrix on a plane and evaluating the reliability of the result= ( H i j ) is defined by IlHll = d m . ing mapping. The basic principle is maximum likelihood We regard data points as random variables. Let estimation by a technique called renormalization. X, and Xb, be the covariance matrices of points (x,, y,) and (xb,, y;), respectively. The covariance