A swarm intelligence based approach for image feature extraction

Abstract

This paper presents a new approach to detect points of interest in an image. It uses swarm intelligence to detect centers of objects which are considered as points of high interest because many of psychological works state that the symmetry attracts the attention of human visual system. This fact led to the choice of symmetric objects' centers as points having a high visual interest and then used as points of interest in object classification and recognition. Earlier works search for points of interest in high signal variations. Unfortunately, with images presenting very limited signal variations, existing approaches seem to be useless. In contrast, the use of approaches dealing with visual interpretability like centers of symmetric objects is suggested to be an interesting research axis. The originality of this work is the use of a nt colonies for recognition purpose whereas the major use of ant colonies is to do optimization and classification.