This paper reviews the basic techniques of motion information recovery from image sequences. It highlights the recursive method of estimating motion parameters as they are computationally less complex than nonrecursive ones. This paper focuses. on the use of an extended Kalman filter (EKF) in 2-D image processing. A straightforward extension of one-dimentional Kalman filter techniques would result in a number of state variables proportional to n for the filtering of an rum digital image. To overcome this problem, some approximations have been used to reduce the computational load of the algorithm but at the expense of its accuracy. This paper investigates also the parallel implementaion of the EKF in 2-D image processing problem for a range of transputer topologies to enable practical realizations. The parallel implementation results are presented and the relative speedup and efficiency measures are calculated to evaluate the perforemance.