Author: Jacob Laurel

Departments:
Department of Electrical and Computer Engineering, University of Alabama at Birmingham, Birmingham, AL, USA
Electrical and Computer Engineering Departmental Honors Program, University of Alabama at Birmingham, AL, USA

Abstract

A novel adaptive median filter is presented that can restore images corrupted by salt and pepper noise levels greater than 90%. The algorithm operates by adapting to the amount of available visual data in the image by iteratively increasing the size of the median kernel. The algorithm then detects the edges and reruns the adaptive median filtering process on just those edge pixels to improve edge consistency. Lastly, post-processing is done on the image using the Perona-Malik diffusion process for smoothing and an Unsharpen filter to improve contrast. The results of our algorithm show root-means-quare error improvement of the reconstruction compared to the state-of-the-art filter for image reconstruction.

Download the full article (PDF): An Adaptive Kernel-Growing Median Filter for High Noise Images