Progressive Transformation: A Simple, Customized Approach for Dimensional Reduction and Scalar Parameter Generation for Remotely-Sensed Data Analysis

Abstract

The extraction of measures of certain properties of objects from multichannel data is a basic process in analytical remote sensing. One approach is the use of standard rotational transformations, e.g., those designed for use in agricultural areas for “soil brightness,” (vegetative) “greenness,” (vegetative) “yellowness,” (plant or soil) “wetness,” etc. from Landsat Multispectral Scanner System data [l] or Landsat Thematic Mapper data [2]. While standard rotational coefficients may work well in many cases, a need exists for a simple approach which will produce a customized set of transformation coefficients which will yield scalar measures of certain physical properties in particular situations. This paper presents, in a tutorial fashion, a technique called the Progressive Transformation for achieving this goal. Key Words: Information extraction, customized transformations, multichannel remote-sensing.