摘要

This paper proposes a novel scheme for texture segmentation and representation based on Ant Colony Optimization (ACO). Texture segmentation and texture characteristic expression are two important areas in image pattern recognition. Nevertheless, until now, how to find an effective way for accomplishing these tasks is still a major challenge in practical applications such as iris image processing. We propose a framework for ACO based image processing methods. Considering the specific characteristics of various tasks, such a framework possesses the flexibility of only defining different criteria for ant behavior correspondingly. By defining different kinds of direction probability and movement difficulty for artificial ants, an ACO based image segmentation algorithm and a texture representation method are then presented for automatic iris image processing. Experimental results demonstrated that the ACO based image processing methods are competitive and quite promising, with excellent effectiveness and practicability especially for images with complex local texture situations.