Study Of Remote Sensor Spectral Responses And Data Processing Algorithms For Feature Classification

Abstract

A computational model of the deterministic and stochastic processes involved in remote sensing is used to study and compare the performance of sensor spectral responses and data processing algorithms for classifying spectral features. The simulated spectral responses include those of the U.S. Landsat Thematic Mapper (TM) and the French Systeme Probatoire d'Observation de la Terre (SPOT). The simulated data processing algorithms include the computationally simple boundary approximation method (BAM) to discriminate between general target categories such as vegetation, bare land, water, snow, and clouds, and the maximum likelihood (MLH) and mean-square distance (MSD) classifications to discriminate between specific targets such as various crop types.