Signature Separability Analyses for Wetland Vegetation Mapping Over Blackwater Wildlife Refuge

Figure 37. Unsupervised classification results of IKONOS image after application of the nine-pixel minimum mapping unit filter

Figure 39. Spectral plots of low reflectance signatures used to classify the southern mosaic

ERDC-TR-05-1
Page Navigation
66

Figure 38. Southern Mosaic--flightlines BLINE7E through BLINE20E. (top)

false-color composite, and (bottom) true-false-color composite.

5.3.2 Signature Separability Analyses for Wetland Vegetation Mapping Over

Blackwater Wildlife Refuge

Before applying the newly created spectral signatures as training data to the

14-band images, the signatures were evaluated for separability. Signature sepa-

rability is a standard tool in commercial image processing software packages. A

variety of algorithms are generally available. ENVI offers two: Jeffries-Matusita

and Transformed Divergence. Both use the n-dimensional mean vectors and co-

variance matrices of the training statistics to measure the magnitude of the differ-

ence, or separability, of the mean vectors. The Transformed Divergence statistics

Western Governors University

Privacy Statement
Press Release
Contact

Integrated Publishing, Inc. - A (SDVOSB) Service Disabled Veteran Owned Small Business