Iterative truncated linear filter for image noise reduction

Abstract

The arithmetic mean and the order statistical median filters are two widely used operations in signal and image processing. Both of them have some merits and limitations in noise attenuation and image structure preservation[1]. This project aims to study the properties of iterative truncated mean (ITM) filter which shows some merits of both the fundamental operations, it is able to estimate the median by simple arithmetic computing. This algorithm truncates the extreme values of samples in the filter window to a dynamic threshold, the dynamic truncation thresholds can guarantee the filter output, starting from the mean, to approach the median of the input samples[1].In this project, Matlab and C programming are used to implement the ITM filters, and Matlab programs are used to test their performance. ITM filter, FITM (fast realization) filter and NITM (new method) filter are tested under different conditions such as different noise distribution. Their properties are analyzed and experimentally verified on synthetic data.