Date of Award

1992

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Geography and Anthropology

First Advisor

Nina Lam

Abstract

This study examines the relationship between resolution and fractal dimensions of remotely sensed images. Based on the results of testing for the reliability of the algorithms on hypothetical surfaces, the isarithm algorithm is selected for determining the fractal dimensions of remotely sensed images. This algorithm is then applied to simulated fractal Brownian motion images and four calibrated airborne multispectral remotely sensed image data sets with different true and artificial resolutions for Puerto Rico. The results from applying the fractal method to images at different levels of resolution suggest that the higher the resolution of an image, the higher the fractal dimension of the image and the more complex the image surface. This relationship between resolution and fractal dimension is further verified by results from analysis employing the local variance method for the same data sets; where it is found that the higher the resolution, the higher the local variance or the more complex the image surface. The images with artificial resolutions were found to be unrealistic in simulating images with different resolutions because the aggregate method used in generating these images dose not exactly simulate the sensor's response to resolution changes. The aggregate method has been widely used in image resampling and cautious use of this algorithm is suggested in future studies. The findings show that the fractal method is a useful tool in detecting the scale and resolution effects of remotely sensed images and in evaluating the trade-offs between data volume and data accuracy. More studies employing fractals and other spatial statistics to images with different artificial resolutions generated using better aggregation algorithms are needed in the future in order to further detect the scale and resolution effects in remote sensing and GIS.

Pages

222

Share

COinS