Horn & Schunck Meets a Discrete Zhang Neural Networks for Computing 2D Optical Flow

Abstract

In this paper, we develops a new fast on line algorithms for motion estimation. This new algorithm is based on the Horn & Schunck algorithm with a new kind of recurrent neural network called Discrete Zhang Neural Networks (DZNN) and Simoncelli’s matched-pair 5 tap filters. We simulate the network on a sequential processor and compare its performance with a sequential algorithm based on the Jacobi method. Experimental results on synthetic and real image sequences for the method are given to demonstrate its fastness in comparison with Jacobi method.