Multiple Textured Objects Segmentation Using DWT Based Texture Features in Geodesic Active Contour

Abstract

This paper addresses the issue of segmentation of multiple textured objects in presence of background texture using active contours. It first comments on the parametric and geodesic active contours in view of their abilities to handle multiple objects segmentation, in general and then proposes a technique based on geodesic active contours for the multiple textured objects segmentation. The proposed approach is based on the generalization of the geodesic active contour from one dimensional intensity based feature space to multidimensional texture based feature space. The main contributions of this work are as follows. First, it develops new scalogram based texture features which have good texture discriminating power and second, use of these features to define a good "texture edge indicator" function to be used in geodesic active contour algorithm. Proposed technique is validated using a set of synthetic and natural texture images. To show the effectiveness of the proposed technique, obtained results are compared with other techniques suggested in the literature.