ADAPTIVE L-FILTERS WITH APPLICATIONS IN SIGNAL AND IMAGE-PROCESSING

Adaptive order statistic filters (L-filters) based on local statistics of signals are introduced in this paper. Such estimators are particul arly effective in filtering signals which are corrupted by more than o ne type of noise, e.g., additive Gaussian noise plus impulses. In orde r to further improve the filtering performance (detail preserving abil ity), adaptive L-filtering structures have been combined with weighted median operations, exploiting temporal information which is lost by o rdering in L-filtering. Simulation results are included to assess the performance of the proposed adaptive structures for one and two dimens ional signals.