Stratification of Landsat data by clusteringFull realization of the potential advantages of the synoptic coverage provided by Landsat will require the development and use of data analysis techniques which take into account the large variation and diversity of patterns found over many Landsat scenes. Stratification of the scene into units which are internally homogeneous is recommended as a first step in the analysis of data for whole or multiple frames of Landsat data. The use of clustering as an objective and efficient method of dividing scenes into areas which are spectrally similar (strata) is discussed and initial results, including classification performances and comparisons of spectral strata with major physical factors, are presented.
Document ID
19770032212
Acquisition Source
Legacy CDMS
Document Type
Conference Proceedings
Authors
Bauer, M. E. (Purdue Univ. West Lafayette, IN, United States)
Davis, B. J. (Purdue University West Lafayette, Ind., United States)
Date Acquired
August 9, 2013
Publication Date
January 1, 1976
Subject Category
Earth Resources And Remote Sensing
Meeting Information
Meeting: Symposium on Machine Processing of Remotely Sensed Data