The Study of The Bay of Mount Saint-Michel by Using Graph Theory in The Analysis of Satellite Images

In this paper, a new approach for mapping based on the concept of objects and relationships between these objects is proposed to take advantage from both supervised and unsupervased classification methods. On the one hand, objects obtained after a supervised classification are represented by an adjacency graph model. On the other hand, objects obtained after unsupervised classification are represented by an adjacency graph data, and the goal is to measure the matching between this two graphs in order to improve the results of unsupervised classification in association with those obtained from supervised classification. This study concerned the coastal Bay of Mont Saint-Michel, the data used are from SPOT 5 optical satellite images.