Fast full-search block-matching algorithm for motion-compensated video compression

Abstract

This paper proposes a fast block-matching algorithm that uses three fast matching error measures, besides the conventional mean absolute error (MAE) or mean square error (MSE). An incoming reference block in the current frame is compared to candidate blocks within the search window using multiple matching criteria. These three fast matching error measures are established from the integral projections, taking their advantages of good representation for block features and simple complexity in measuring matching errors. Most of the candidate blocks can be rejected by calculating only one or more of the three fast matching error measures. The time-consuming computations of MSE or MAE are performed on only a few candidate blocks that first pass all the three fast matching criteria. Simulation results show a reduction of over 86% in computations is achieved after integrating the fast three matching criteria into the full-search algorithm, while ensuring the optimal accuracy.