A Technique Utilising a Combination of Supervised and Unsupervised Classifications in Solving Geological Tasks

Remote sensing has found many geological applications. The main problem in solving geological tasks is spectrum mixing. We propose the remote sensing multispectral data interpretation method which decreases the effect of spectral mixing without applying mathematical models for spectrum division. The main feature is successive using of unsupervised and supervised classification. The method was applied in task of investigating lode gold deposits.