Hyperspectral Imaging Based Method to Identify Potential Limestone Deposits

Surface mineral deposit identification using Hyperspectral imaging is important since it is economical. In this paper, an algorithm to detect potential limestone deposits is proposed with the use of a self-generated limestone reference. For this task images obtained by EO-1 satellite's Hyperion sensor are used. In the proposed algorithm, first, Principal Component Analysis is applied on the dataset for dimensionality reduction and a Euclidean distance based classification to identify pixels corresponding to soil regions. Next, Correlation Factor Analysis and Fisher Discriminant Analysis (FDA) are used to classify the regions based on their limestone availability. Finally, a reference signal is generated from the dataset to use with FDA to improve the accuracy of the results.