IMPLEMENTATION OF FUZZY FILTERS FOR SMOOTHING & SHARPENING OF DIGITAL IMAGES

Abstract

Image smoothing is an important step in many image-processing applications. In the presence of noise, smoothing is an important preprocessing step if they are to be followed by other tasks such as edge detection, feature extraction and object / pattern recognition. The most effective approaches of image smoothing are non-linear and adaptive in nature. Noise smoothing and edge sharpening are inherently conflicting processes, since smoothing a region might destroy an edge and sharpening edges might lead to unnecessary noise.The type of algorithm to be used depends upon the objective to be achieved by the smoothing process as well as the particular application. The objective of the thesis is development of fuzzy filters, which are adaptive in nature and remove different noises like salt and pepper and Gaussian noise. Most of the classical filters that remove noise simultaneously also blur the edges, but fuzzy filters have the ability to combine edge-preservation and smoothing. Compared to other non-linear techniques, fuzzy filter are able to represent knowledge in a comprehensible way.