An error analysis of camera translation direction estimation from optical flow using linear constraints

A linear method using optical flow to recover the translation direction of a moving camera has been previously presented. This algorithm produced a biased estimate and various compensation techniques have been proposed. The results of an error analysis of the different approaches containing expressions for the expected mean vectors and covariance matrices of the translation estimates are presented. This analysis includes a necessary modification to the Cramer-Rao lower bound which specifies the minimum possible covariance matrix for the different estimators. The analytical results were verified experimentally using simulations and real images.