A radiometric normalization technique for compensating illumination and atmospheric differences between multi-temporal images should allow classification of the images with a single classification algorithm. This allows a simpler approach to land cover change detection. Land cover classification of Landsat Thematic Mapper Imagery with and without Pseudo Invariant Feature Normalization was performed to demonstrate the effect on classification and change detection accuracy. A post-classification change detection method using two separate classification algorithms, one for each date, was performed as a baseline comparison. Land cover classification using one classification algorithm was attempted with and without gain and offset correction to serve as another comparison. Accuracy verification was performed on the classification results by comparing random samples against ground truth.
Library of Congress Subject Headings
Landsat satellites; Artificial satellites in remote sensing; Remote sensing--Mathematics; Electromagnetic waves
Department, Program, or Center
Chester F. Carlson Center for Imaging Science (COS)
Hawes, Tim, "Land cover classification of landsat thematic mapper images using pseudo invariant feature normalization applied to change detection" (1987). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus