Nonlinear signal and image processing based on order statistics

With the emerging developments in all areas of telecommunications and multimedia technology, the need for fast, effective, and robust filtering schemes over noisy channels has increased. In this dissertation, new techniques for enhancing the speed, effectiveness, robustness, design simplicity, and adaptiveness of the existing filtering technology are presented by developing innovative techniques for the design and implementation of filters based on order statistics (order filters). Since the proposed filter types utilize a pure nonlinear operation (ordering) and a pure linear operation (weighting), they show exceptionally good performance in removal of the noise types that cannot be eliminated by conventional linear filters. By overcoming the existing problems of order filters, the new techniques allow filter implementation in the form of real-time adaptive nonlinear filters.