Adaptive Order-Statistic LMS Filters

Abstract

The LMS-based adaptive order-statistic filters are presented in this paper. The adaptive Ll-filters as extension of the adaptive L-filter for two-dimensional filtering of noisy greyscale images is studied too. Their adaptation properties are studied by three types of noise, the additive white Gaussian noise, the impulsive noise or both, respectively. Moreover, the impulsive noise has the fixed noise value (Salt & Pepper noise). The problem of pixel value multiplicity and determination its position in the ordered input vector for adaptive Ll-filter is shown in this article. The two types of images with different of image complexity are used to demonstration of the power of time-spatial ordering.