Image noise reduction by iteratively truncated mean filter

Abstract

It is well known that noise filtering is very important in image and signal processing. Common noise filters are mean and median filters which they have respective pros and cons in noise attenuation and image structure preservation. Therefore there is a necessity and tendency to develop another type of noise filter to combine the merits of both mean and median filters. Iterative Truncated Arithmetic Mean (ITM) filter is introduced and developed to have the merits of both mean and median filters, which starts from the mean to approach the median. Another type of filter Fast Iterative Truncated Arithmetic Mean (FITM) filter based on ITM filter is also proposed to be expected to have faster running speed compared to ITM filter. Both ITM and FITM filter will be implemented by MATLAB and C code respectively. A lot of experiment will be conducted as well to compare the results of two different implementation methods, the performance compared to median filter as well as the running speed.