Feature Extraction Approach in Hyperspectral Data

In general, feature extraction deals with the problem of finding the most informative, distinctive, and reduced set of features and improve the success of data processing. The features should contain information required to distinguish between classes, be insensitive to irrelevant variability in the input, and also be limited in number, to permit, efficient computation of the applied functions and to limit the amount of data required. In many cases, it is an important step in the solutions of many tasks aiming to extract the relevant information from the available large datasets. The aim of this study is to apply a feature extraction approach to a hyperspectral image and extract different features from the dataset and reduce its dimensionality into meaningful orthogonal features. The final analysis was performed in a test site situated in central Mongolia using 242 band Hyperion data. Overall, the study indicated that the Hyperion hyperspectral data could be effectively reduced into meaningful features through a feature extraction process.