Multichannel nonlinear filters for signal restoration

Abstract

Multichannel nonlinear filtering techniques are becoming increasingly important in signal processing applications. In general they perform better than using only one channel. Furthermore, nonlinear filters are often better than linear filters at removing noise without distorting signal features. The main focus in this dissertation is comparing different class of nonlinear multichannel filters. In this paper, a new class of nonlinear filters is introduced and analysed, this filter is the hybrid, (H) filters. These filters are formed by taking a linear combination of both the observation samples in temporal or spatial orders and the sorted observed samples. This work also extends the single channel H filters and LI filters to the multichannel case. The definitions of the filters are given and compared to the other multichannel filters. The behavior of these filters are studied for a multichannel signal with contaminated additive Gaussian noise. The estimated signal is optimum under the criterion of the Mean Square Error (MSE) between the output of the filter and the signal or image being estimated. Experimental results show that the proposed filters produce lower MSE than many well-known filters.