Image Segmentation by Self-Organised Region Growing

In this paper, our work aims to show how region growing technique may be handled by swarming agents that are self-organisation in response to the local conditions of an image environment. In this approach, we combine the stigmergy mechanism with another type of swarm behavior: particle swarm optimization (PSO). PSO is used for optimize the search of the seed pixels of homogeneous regions and adjust locally the homogeneity threshold in different parts of image. However stigmergy is used for regulation of the growing structure of homogenous regions. Some experiments conducted on synthetic and real gray level images for showing the features of this approach and to present the obtained results.