arXiv papers

Download open scientific papers. Check out the SCG's arXiv papers

A New Approach to Estimate Fractal Dimension of Texture Images

Andre Ricardo Backes and Odemir Martinez Bruno

International Conference on Image and Signal Processing,136-143,2008

One of the most important visual attributes for image analysis and pattern recognition is the texture. Its analysis allows to describe and identify different regions in the image through pixel organization, performing a better image description and classification. This paper presents a novel approach for texture analysis, based on calculation of the fractal dimension of binary images generated from a texture, using different threshold values. The proposed approach performs a complexity analysis as the threshold values changes, producing a texture signature which is able to characterize efficiently different texture classes. The paper illustrates the novel method performance on an experiment using Brodatz images.