Can crop types be resolved using mixture distribution components - Some initial results and implicationsFor the analysis of remotely sensed data, it is frequently necessary to design a classifier in order to locate a ground cover class of interest or to estimate the proportion of this ground cover class. Advantages of a mixture distribution formulation are discussed, and a description is presented of the results of estimating the proportion of small grains in ten Landsat data segments using the mixture model. It is found that the mixture model proportion estimates have a very low variance and coefficient of variation. The discussed investigation implies that the mixtures model is a viable method for determining the distributions of classes of interest in remote sensing problems and in estimating the proportions of these classes directly.
Document ID
19840030278
Acquisition Source
Legacy CDMS
Document Type
Conference Paper
Authors
Lennington, R. K. (Lockheed Engineering and Management Services Co., Inc. Houston, TX, United States)
Sorensen, C. T. (Lockheed Engineering and Management Services Co., Inc. Houston, TX, United States)
Heydorn, R. P. (NASA Johnson Space Center Houston, TX, United States)