An Application of Neuro-fuzzy System in Remote Sensing Image Classification

This paper introduces a classification algorithm-NEFCLASS (neuro-fuzzy classification) to classify remote sensing images landsat7 etm+. The NEFCLASS combines neural networks and fuzzy systems to learn from the training data and generate conditional linguistic rules. Then we use the rules to classify the land-cover/land-use classes in landsat7 etm+ images covering main Bayannaoer city of Inner Mongolia Autonomous Region selected in August 2007. Compared to the ground truth, the experiment result shows that the overall classification accuracy can achieve to 79.93%.