Remote Sensing Image Classification Based on Probabilistic Neural Network

This paper has analyzed the remote sensing classification using the probabilistic neural network (PNN) on the basis of data normalization, for the best classification accuracy, the picking of sample area and the standard deviation of the basis Gauss function has been discussed.PNN classification model was applied to classify the TM image in Tibet.Based on error matrix, the classification result of the maximum likelihood was contrasted with that of PNN model.The results show that the overall accuracy and Kappa coefficient of PNN model reach 94.5% and 0.934, respectively, is superior to that of the traditional Maximum likelihood method.