[en] Analysis of VNIR-SWIR hyperspectral images is presented to assist the development of a multi-sensor scanning system for Iberian Pyrite Belt Cu-Zn-Pb projects. Fisher Linear Discriminant and Linear Support Vector Classifier were used for supervised classification after pre-processing, spectral plotting and construction of false color composites. Validation is given by mean accuracy of confusion matrices for different scenarios considering parameters of practical applications in industrial settings. Interpretation indicates a different performance for shale and volcanic-hosted deposits. The results demonstrate the power of machine learning algorithms and hyperspectral databases applied to an automated technique to assist the traditional logging. Combined to other sensors, the methodology should be adapted to a drill-core scan in mining projects, delivering efficient and time-saving outcomes.
