Discrimination of Satellite Signals from Opencast Mining of Mineral Ores of Hematite and Uranium Using Digital Image Processing and Geostatistical Algorithms

Abstract

This study has been conducted on mining areas of Gua and Banduhurang in Jharkhand State, India, Barbil in Orissa State, India and Mines d’Arlit, Niger. Data from the corresponding sensors of EO-I, LANDSAT and IRS (i.e., Hyperion, ETM+ and LISS III respectively) were used to discriminate satellite signals of Hematite and Uranium ores from these locations. For the data of Hyperion being hyper-spectral, correction mechanism were performed through relevant algorithms: Fast Line-of-sight Atmospheric Analysis of Hypercube (FLAASH) to remove atmospheric aerosol effects, Minimum Noise Fraction (MNF) to remove noise, Pixel Purity Index (PPI) to get spectrally the most pure pixel and Spectral Angle Mapper (SAM) in order to match the spectral similarity between an image pixel spectrum and a referenced spectrum. The data achieved from ETM+ had line-stripping, and thus were restored. On the LISS III data, vegetation had to be, virtually, removed from the images of the Indian sites, using Normalized Difference Vegetation Index (NDVI), in order to equate them with that from the Niger site. The processed data was put to a common platform statistically. Segregation of Uranium, a radioactive ore, from Hematite, a non-radioactive iron ore, could be achieved up to 82.35% using TOPSIS and 90% using pair-wise Student’s t-Test. The technique of Band Ratio was also carried out and an index was generated to isolate these mines from their surroundings.