IMPROVED MULTISCALE EDGE DETECTION OF NOISY DIGITAL IMAGES USING DWT AND ITS PERFORMANCE ESTIMATION

Abstract

The wavelet transforms has been perhaps the most exciting development during the last decade to bring together researchers in several fields such as image processing, signal processing, communications, computer science and mathematics. Less obvious; but becoming increasingly well known, is that signal and image processing especially are getting great improvements in performance by using wavelet based methods. The beginnings of the wavelet transforms as a specialized field can be traced back to the works of Grossman and Morlet (1984) for which the motivation was drawn from the study of the fact that certain seismic signals can be modeled suitably by combining translations and dilations of a simple oscillatory function of finite duration called wavelet [3]. The ability of the wavelet -transform ' to localize events both in time and frequency is the, single important factor that makes it more .:useful in processing any kind of data. Application of the wavelet transform has almost come to be regarded as being synonymous with data compression and thereby applicable for image compression as well. However, there are other properties of the wavelet transform that make it naturally suited for application in many other areas of digital image processing. An effort has been made. in this dissertation, work to achieve certain degree of improvement in the performance of one such wavelet application ie. Multiscale image edge detection, especially when images are clouded by noise of different intensities.