You are being redirected because this document is part of your ASTM Compass® subscription.
    This document is part of your ASTM Compass® subscription.


    Application of Spatial Statistics to Analyzing Multiple Remote Sensing Data Sets

    Published: 0

      Format Pages Price  
    PDF (456K) 13 $25   ADD TO CART
    Complete Source PDF (7.2M) 275 $65   ADD TO CART


    The purpose of this paper is to present results from preliminary research using image spatial statistics to improve the resolution of low-resolution digital images. Improvement is needed whenever images from different sensors, and often when images from different spectral bands within the same sensor, are collected for geotechnical interpretation.

    The experimental procedure involves degrading one of two digital images in adjacent spectral bands to simulate a low-resolution image. New samples within the sparse, low-resolution image are subsequently estimated using the spatial characteristics displayed in a variogram function and an estimation process known as co-kriging or co-estimation. The resulting high-resolution estimation image is compared with the original undegraded image as a check on estimation accuracy. Reconstruction accuracy, as judged by mean square error (MSE), improved from MSE = 826 for sample replication to 56 for co-kriging.

    For cases where geotechnical decisions rely on accurate knowledge of image amplitudes, this research suggests that co-kriging provides an accurate and flexible estimation technique. Moreover, co-kriging appears to be a promising technique for future automated image interpretation and feature extraction algorithms.


    remote sensing, image processing, geostatistics, image registration

    Author Information:

    Glass, CE
    The University of Arizona, Tucson, AZ

    Carr, JR
    University of Nevada, Reno, NE

    Yang, H-M
    The University of Arizona, Tucson, AZ

    Myers, DE
    The University of Arizona, Tucson, AZ

    Committee/Subcommittee: D18.01

    DOI: 10.1520/STP25979S