Abstract
Most mining companies have registered important amounts of drill core composite spectra using different acquisition equipment and by following diverse protocols. These companies have used classic spectrography based on the detection of absorption features to perform semi-quantitative mineralogy. This methodology requires ideal laboratory conditions in order to obtain normalized spectra to compare. However, the inherent variability of spectral features—due to environmental conditions and geological context, among others—is unavoidable and needs to be managed. This work presents a novel methodology for geometallurgical sample characterization consisting of a heterogeneous, multi-pixel processing pipeline which addresses the effects of ambient conditions and geological context variability to estimate critical geological and geometallurgical variables. It relies on the assumptions that the acquisition of hyperspectral images is an inherently stochastic process and that ore sample information is deployed in the whole spectrum. The proposed framework is basically composed of: (a) a new hyperspectral image segmentation algorithm, (b) a preserving-information dimensionality reduction scheme and (c) a stochastic hierarchical regression model. A set of experiments considering white reference spectral characterization and geometallurgical variable estimation is presented to show promising results for the proposed approach.
Highlights
Mining industry has taken advantage of near infrared (NIR) spectrography [1] technology to characterize geological and metallurgical samples
We propose a stochastic characterization method which consists of process pipelines to train and predict/estimate interest variables from metallurgical sample hyperspectral images
simple linear iterative clustering method (SLIC) method is an image segmentation strategy, based on a local spatial partition clustering approach, that has been widely adopted in applications involving common digital images, because it has shown: (a) a good performance, (b) to be fast at finding object borders and (c) a linear convergence time proportional to the number of target super-pixels and to the number of pixels contained in the image
Summary
Mining industry has taken advantage of near infrared (NIR) spectrography [1] technology to characterize geological and metallurgical samples They record drill core trays using hyperspectral single pixel scanners, such as Hylogger [2], or 2D imaging systems such as Corescan [3,4], and they acquire spectra from composites of drill core segments using single pixel hyperspectral instruments such as ASD SpecLab [5,6,7]. Spectra are used mainly for clays and certain mineral identification tasks, based on classical NIR analysis [1,10] This large amount of available information has produced relevant challenges and difficulties, which gave us our initial motivation to work firstly with machine learning approaches, with some success for geological sample analysis applications, and more recently with a novel stochastic approach for metallurgical sample analysis applications. Classical NIR-based spectrography methods [1], such as characterization of spectral absorption features and unmixing techniques [9], were the first approaches to be used, and a number of standardized methodologies were developed for mineral classification over geological samples
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
