Improving hyperspectral image classification accuracy using Iterative SVM with spatial-spectral information

Hyper-Spectral Images (HSI) classification is one of essential problems in hyperspectral image processing and one of the major difficulties in supervised hyperspectral image classification is the limited availability of training data, as it is hard to obtain in real remote sensing scenarios. In this paper we have presented our proposed approach to improve the accuracy of HSI in the situations where the training samples are very limited and also where we attain misclassification due to random training samples. Our proposed approach is based on the Iterative Support Vector Machine (ISVM) and also on the spatial and spectral information. In order to improve the performance of ISVM, the Majority Voting (MV) and the marker map correction techniques are used to correct the training samples at each iteration of ISVM. Experiments on practical Hyperspectral images including AVIRIS Indian Pine Image are conducted and the results shown that the proposed approach works better than ISVM and other classifiers such as SVM-RBF, Linear-SVM and K-NN.