Construction of a New Classifier Integrated Multiple Sources and Multi-temporal Remote Sensing Data for wetlands

摘要

This paper developed a new classifier named extension model of fuzzy matter-element to classify multi-source and multi-temporal remote sensing data. Taking the National Nature Reserve of Ruoergai wetlands as study area, this paper chose the data of TM, CBERS and MODIS-NDVT of 2007 as data source and introduced the constructing process of this classifier. Results showed that, the overall accuracy (82.36%) and Kappa coefficient (0.8006) of the integrated of multi-information classification method is better than the SVM results using on TM(79.74%, 0.7704) and on CBERS (77.29%, 0.7436). Meanwhile, the paper also made an active exploration of classification on the cloudy image.