Image Segmentation Using Ant System-BasedClustering Algorithm

摘要

Industrial applications of computer vision sometimes require detection ofatypical objects that occur as small groups of pixels in digital images. These objectsare difficult to single out because they are small and randomly distributed. In thiswork we propose an image segmentation method using the novel Ant System-basedClustering Algorithm (ASCA ). ASCA models the foraging behaviour of ants, whichmove through the data space searching for high data-density regions, and leavepheromone trails on their path. The pheromone map is used to identify the exactnumber of clusters, and assign the pixels to these clusters using the pheromone gra-dient. We applied ASCA to detection of microcalcifications in digital mammogramsand compared its performance with state-of-the-art clustering algorithms such as 1DSelf-Organizing Map, k-Means, Fuzzy c-Means and Possibilistic Fuzzy c-Means.The main advantage of ASCA is that the number of clusters needs not to be knowna priori. The experimental results show that ASCA is more efficient than the otheralgorithms in detecting small clusters of atypical data.