Q-L transformations for adaptive generalized order statistic filtering

The authors introduce quasi-range based (Q) transformations in the time, rank and time-rank domains. The eigenanalysis for each mapping is performed, showing that one can enhance the performance of gradient-based adaptive filters in all of these signal spaces by exploiting the improved eigenstructures attained by the transformations. Thus, for adaptive signal processing, in general it is better to use quasi-ranges rather than order statistics