Classification Using EO-1 Hyperion Hyperspectral and ETM Data

Hyperspectral remote sensing technology demonstrates the capacity for accurate vegetation identification. Hyperion image acquired simultaneously in 356-2577nm narrow spectral bands with Wnm sampling interval allows the capture of vegetation types information. This paper employed hyperion hyperspectral data and ETM+ data acquired over Xishuangbanna tropical vegetation area in Yunnan province, China to make agriculture land classification. Remote sensing data geometric correction and radiometric correction preprocessing were first made. Atmospheric correction of the Hyperion image was performed using second simulation of the satellite signal in the solar spectrum (6S) radiation transfer model. The first three principle components of hyperion accounted for 99.3% of total information, which were selected to make unsupervised classification using ISODATA algorithm. Overall classification accuracy of hyperion using spectral angle mapper (SAM) and unsupervised classification after principle component analysis were 52.3% and 65.7% respectively.