Abstract:

For the large-scale remote sensing imagery which has large data and local feature heterogeneity, this paper presents a color remote sensing imagery segmentation method based FCM within a multi-agent system framework. Firstly, dividing the image domain into several sub-domains with rule division tecnology and each segmentation agent controls one sub-domain. And then, the segmentation agent executes the segmentation of its sub-domain by FCM to get the initial local segmentation result within MAS framework, via the harmonizing the local result with the global segmentation result to obtain the true local class number and unify the local label. Finally, the coordinate agent is employed to eliminate the local feature heterogeneity of remote sensing imagery via coordinating the cluster centers of each segmentation agent with its neighbors. The experiments are carried out on the synthetic imagery and the color remote sensing images. The results are qualitatively and quantitatively evaluated and certify the efficiency and effectiveness of the proposed algorithm.

Key words: multi-agent system, neighbor coordinate, image segmentation, FCM, color high-resolution remote sensing imagery