Active contour texture segmentation in modulus wavelet feature spaces

Abstract

In this paper we discuss a model that is able to segment textures using active contours. Our technique is based on active contour techniques using curve evolution. We build our model on properties of human vision, in that we segment the textures in a certain feature space. We will show the advantages of using modulus feature spaces. Wavelet coefficients are shown to exhibit local features both in space and frequency domains. We will implement our model in modulus wavelet subbands. © 2013 Springer Science+Business Media.