Analysis of Multi-Frequency Polarimetric SAR Data Using Different Classification Techniques

Classification of polarimetric SAR images has become a very important topic after the availability of Polarimetric SAR images through different sensors like SIR-C, ALOS-PALSAR etc. The data over wet regions of India has been processed for classification of various land features like mangrove, ocean water, and clear water. In this study the utility of NASA’s Shuttle Imaging Radar-C (SIR-C) data is evaluated for wet regions of India. Supervised and unsupervised classification techniques are used to classify the data. The SIR-C data is acquired over Kolkata region of West Bengal, India. The results show that multipolarization and multi-frequency SAR data helps to classify wetlands effectively. The combinations of different polarizations from Land Cband helps to improve the classification accuracy. It was found that the combinations of channels (L-HV, C-HH, C-HV) and (L-HH, C-HH, C-HV) gave the best overall accuracies. These two 3 channel combination can differentiate well the six classes. The five band combination L-HH, L-HV, L-VV, CHH, C-HV gives the highest classification accuracy. It is greater than the three band combinations as given above. By applying enhanced Lee filter the accuracy can be further increased. The enhanced Lee filter removes the speckle effectively. Among all the classifiers Maximum Likelihood classifier gives the best accuracy.