Unsupervised Change Detection Using Multi_temporal SAR Images Based on Improvement of Level Set Methods

In this research, a framework is presented for the unsupervised change detection using Multitemporal SAR images based on integration of Clustering and Level Set Methods. Spatial correlation between pixels is considered by using contextual information. Furthermore, in the proposed method integration of Gustafson-Kessel clustering techniques (GKC) and Level Set Methods are used for change detection. Using clustering techniques has caused production of the initial curve for LSM with a minimum convergence time and, as a result, use ofLSMLeads to increasing the accuracy of change map using the iterative process. In order toevaluate theperformance of the proposed method, this method is compared with some other existing stateof- the-art methods. The results show that the total error rate of the proposed method has been reduced compared to these methods. Results show the high capability of the proposed method in the unsupervised change detection of multi-temporal satellite SAR images.