Abstract

   The subject of this article is to present an efficient method of image segmentation based on a nature inspired optimization technique namely Particle Swarm Optimization and a classification of the segmented images with support vector machines (SVM linear classifier .The classification is performed on the segmented image itself in one case and on the segmented image pre-processed by applying a Karhunen-Loeve transformation that de-correlates and compress the image in the other case.