WSEAS Transactions on Biology and Biomedicine
Print ISSN: 1109-9518, E-ISSN: 2224-2902
Volume 13, 2016
Application of Fractal and Multifractal Analysis Algorithms to Image Segmentation and Classification
Authors: ,
Abstract: Segmentation is one of image processing methods that results in partitioning a digital image into multiple parts (segments). This method may be used as a step in image analysis or may be considered as a single task. Segmentation is used to analyze medical images, faces, images of the Earth surface and many others. For complex structured images researchers in increasing frequency have to elaborate new methods of segmentation. In particular the application of classical segmentation methods to high resolution images often is not very successful, and fractal and multifractal methods seems to be more preferable, because they are more sensitive to pixel intensity changes. In this work we consider two methods of segmentation: calculation of fractal signatures for the elements of a partition of the surface formed by the image grey level function, and the method of multifractal spectrum calculation based on using density function. The both methods result in obtaining a set of numerical characteristics that may be used as classification and segmentation signs. The both methods show reliable results for images from Brodatz album and high resolution medical images of various types.
Search Articles
Keywords: Fractal dimensions, fractal signature, multifractal spectrum, image segmentation and classification, density function, level sets
Pages: 14-21
WSEAS Transactions on Biology and Biomedicine, ISSN / E-ISSN: 1109-9518 / 2224-2902, Volume 13, 2016, Art. #3