An advanced system for classification of multitemporal SAR images is presented. The system is composed of a feature-extraction module and a neural-network classifier. The feature-extraction module derives a set of features (which are based on long-term coherence and temporal variability) from a series of multitemporal SAR images. The neural-network classifier (which is based on a radial basis function neural architecture) properly exploits the multitemporal features for producing accurate land-cover maps. Experimental results (obtained on a multitemporal series of ERS-1 SAR images) confirm the effectiveness of the proposed system, which exhibits both high classification accuracy and good stability with respect to the architecture of the neural classifier.
展开▼
