Extraction of wetland combing with Radarsat and HJ data of Yellow River Delta

Wetland types of Yellow River Delta are various and serious phenomena of ´same object with different spectrum´ and ´different object with same spectrum´ is one of the reasons caused low classification accuracy. Combination with multi source images is an efficient method to mitigate this influence. In the paper, principal component transform was carried out to Radarsat four polarization data and the first principal component were fused with HJ images based on HIS, Brovey, PC and Wavelet transform. A maximum likelihood classifier was applied to extract wetland information of Yellow River Delta. The experiment results demonstrated that HIS transform performed well than the others and outstood the wetland information. The results also showed that the classification accuracies of HIS merged images and the stacked images were highest, through combining two different source data to make good use of information.