sMAE: An Improved Block Matching Criterion

In this paper a new variation of the well known MAE (Mean Absolute Error) matching criterion for video coding applications, is introduced. The MAE criterion is simpler than the MSE criterion because of the lack of multiplication operations, but it achieves a lower performance in terms of PSNR and MSE. In this work we impose an additional statistical constraint over the MAE thus filling the performance gap between the original MAE and MSE criteria. This is accomplished by multiplying the MAE by the deviation of the mean value of the differences. Simulation results show that the proposed criterion performs always better than the MAE, at the expense of an increased computational complexity.