Abstract
The Indiana State University Remote Sensing Laboratory (ISURSL) recognized that the promise of low-cost earth resource evaluation through machine-assisted processing of multispectral (MS) data has not been fully realized. In response to this problem the ISURSL has developed a complete low-cost system of processing MS data which minimizes analysis time for both man and computer while simultaneously maximizing utilization of the data. The basis of the ISURSL classification algorithm, designated LEVELS CLASSIFIER, is identification of numeric boundaries located in multidimensional feature space which differentiate features of interest. Land use classes of interest to an analyst are described by the range of radiance (relative spectral response) levels which define these boundaries. The identification of levels boundaries which accurately defines an earth surface feature is accomplished through sophisticated single and multidimensional histogram terrain analysis. This approach to multispectral data analysis has been shown to be cost effective and accurate in several applied research projects at ISURSL.
Date of this Version
1977