A combined approach using meta-heuristic and mathematical operators for images segmentation

In this paper we propose a multi-agent approach based on the combination of a meta-heuristic approach with mathematical operators for image segmentation. Our model used resolution, edge and region agents. The role of resolution agent is to describe the contents of the image using multiple representations with decreasing resolution while the edge agent detects edges in one level of resolution images. Region agent makes use of spider´s webs to detect different regions. The proposed approach has been implemented in the JADE agent platform over synthetic and a satellite image. An unsupervised measure was used to improve the effectiveness of our model.