WAVELET DE-NOISING IN IMAGE SEGMENTATION

Abstract: Segmentation techniques include many methods which form a fundamental of image regions recognition and classification. The paper is devoted to the image segmentation using different methods and to comparison of their robustness with respect to various levels of noise. Proposed de-noising procedures are based upon appropriate thresholding of wavelet transform coefficients eva-luated by selected decomposition functions. Resulting methods are verified for simulated images and then applied for selected MR biomedical images containing different structures. Sensitivity of segmentation to image components illumination, scale, translation and rotation is mentioned as well