Accession Number : ADA325081

Title :   Kriging, Optimal Sampling and Filtering by Kriging Analysis.

Descriptive Note : Interim rept. no. 3, May-Jul 96,

Corporate Author : READING UNIV (UNITED KINGDOM)

Personal Author(s) : Oliver, Margaret A. ; Webster, Richard

PDF Url : ADA325081

Report Date : 20 AUG 1996

Pagination or Media Count : 27

Abstract : This is the third report on the second phase of the project to apply geostatistics to remote imagery. The main focus of this period of work has been to apply the geostatistical technique of kriging to the image information for Fort Benning in different ways. The first section describes how the kriging equations can be used to optimize sampling effort. The results of this will relate to the last phase of work for this contract on data compression. They show how the estimation variance increases as the grid spacing increases or the number of data retained decreases. Ordinary kriging was used to map the data for Channels 2, 3 and NDVI to produce contour maps. The aim eventually is to assess how well the categories in these maps reflect the different classes of ground cover. The image data were finally filtered by kriging the long and the short range components in the vegetation. The maps of the short range component show the local rate of change in the pattern of ground cover while those of the long range component show the extent of the major patches of different kinds of ground cover.

Descriptors :   *IMAGE PROCESSING, *STATISTICAL SAMPLES, OPTIMIZATION, DATA MANAGEMENT, RANDOM VARIABLES, ANALYSIS OF VARIANCE, MATHEMATICAL FILTERS, LONG RANGE(DISTANCE), PATTERN RECOGNITION, SHORT RANGE(DISTANCE), MAPPING, DATA COMPRESSION, PIXELS, AREA COVERAGE, SATELLITE IMAGERY, UNITED KINGDOM.

Subject Categories : Statistics and Probability
      Cybernetics

Distribution Statement : APPROVED FOR PUBLIC RELEASE