Automated Class Labeling Of Classified Landsat TM Imagery Using a Hyperion-Generated Hyperspectral Library

摘要

Image classification remains dependent on user intervention for class label assignment. Whether that effort takes place in advance of or post classification is immaterial. This paper explores a novel approach to automating the assignment of class labels using a normalized spectral distance measure and a hyperspectral library. The technique resulted in an automatically labeled agricultural map with an overall classification accuracy of 51 percent. outperforming the manual labeling (40 percent to 45 percent accuracy, depending on the analyst performing the labeling) and the Spectral Angle Mapper classifier (39 percent), and was comparable to, or lower than, the classification accuracy of a Maximum Likelihood supervised technique (53 percent to 63 percent) depending on the analyst. The newly developed class-labeling algorithm provided better results for the majority of targets while having similar performance to manual labeling on targets that are particularly difficult to differentiate in a purely spectral manner