Asian Transactions on Fundamentals of Electronics, Communication & Multimedia

Volume: 02, Issue: 01, March 2012
ISSN 2221-4305

Title: Adaptive Switching Median Filter Utilizing Quantized Window Size to Remove Impulse Noise from Digital Images
Author: Haidi Ibrahim
Paper ID: ATFECM-80211019
Pages: 1 - 6
Abstract: A new impulse noise reduction filter is proposed in this paper. This method, known as Quantized Adaptive Switching Median Filter (QASMF), is derived from Simple Adaptive Median Filter (SAMF). QASMF is constructed from two main processing blocks. In the first processing block, the damaged image is filtered by using adaptive switching median filter, utilizing quantized window size. The quantization of the window size is carried out based on the number of "noise-free" pixels within a local 3x3 neighborhood. The second processing block employs SAMF to filter the residual “noise” pixels from the first processing block. The size of the median filter will expand until there are at least eight "noise-free" pixels contained within the filtering window. Experimental results using five standard grayscale test images of size 512x512 pixels show that QASMF is able to restore corrupted images, even when the images were highly degraded by impulse noise. As compared with Adaptive Switching Median Filter (ASMF) and SAMF, QASMF has the lowest Mean Square Error (MSE) value at any impulse noise corruption level. QASMF requires less processing time than SAMF if the input images were corrupted by 40% or more impulse noise.

Full Text: PDF (756 KB)