Texture segmentation using Shanon wavelet

Abstract

In this paper, we present an approach to texture segmentation that utilizes the Hurst coefficient or the fractal dimension computed along the 1-D cross sections of 2-D texture data. These coefficients are computed utilizing the Shanon wavelet fractal estimation algorithm using a maximum likelihood estimate. These coefficients are considered the feature vector which is used to achieve segmentation using supervised or unsupervised techniques. The major advantage of the Shanon fractal estimator is its simplicity due to the pyramid structure used. The approach has been tested on real brodatz textures and outdoor scenes and yielded the appropriate segmentation.