Non-linear Filters for Removal of Gaussian Noise

Abstract: Images are very often degraded by Gaussian noise during image acquisition. Gaussian noise is evenly distributed over the image and characterized by adding to each image pixel a value from a zero-mean Gaussian distribution. It is necessary to remove Gaussian noise from the image while keeping its features intact. A review of non-linear bilateral filter, trilateral filter, fast and efficient algorithm, modified fuzzy filter and fast guided filter is presented in this paper. This study shows that computational complexity and performance of filters for flat & busy regions and at edges plays vital role in overall assessment of the filter.

Keywords: Gaussian Distribution, Domain, Range, ROAD, Threshold, Weighting Coefficient.