Multiscale MSE-minimizing filters for gradient-based motion estimation

Gradient-based algorithms play a vital role in motion estimation. In this paper, a motion estimation algorithm based on gradient methods for the low signal-to-noise ratio (SNR) scenarios was presented using statistical performance of the estimator. The cost function model of mean square error (MSE) was developed based on the Cramer-Rao lower bound by considering the influence of the noises on motion estimation. The optimal gradient filters for motion estimation were obtained by minimizing the MSE cost function. In combination with the multiscale pyramid approach, the accuracy of such a motion estimation algorithm can be further improved. Experimental simulations show that the proposed method improves the estimator performance for low SNR scenarios.