Comparison of physics based processing to orthogonal subspace projection methods for the classification of vegetation in high resolution hyperspectral data

In October 1997, the TRWIS III sensor was mounted into a small airplane for the purpose of collecting hyperspectral data over a variety of scenes in Ventura County, California, a largely agricultural area about 100 km from Los Angeles. The resulting hyperspectral 384 band data were analyzed using three different approaches. The first was a physically based procedure using the ratios of spectra selected based upon ground truth. Ratios between images in different bands is a way to emphasize the spectral difference and minimize the effect of illumination. The spectral bands selected were in the vicinity of the near infrared "red edge" chlorophyll feature. In the second procedure the original hyperspectral data are transformed using a principal component (PC) projection. Once the data were transposed to this new coordinate system, they were formed into images of the PCs. However, the images from the PCs did not give results separating the tree types unless the spectral region was restricted to 30 bands in the vicinity of the red edge feature. In the third method, the data were processed to find spectral end-members and the hyperspectral data cube spectrally unmixed using these end-members into individual component images. The autonomous end-member determination method was able to differentiate between the tree types when the full visible/near infrared data set was used.