Full text loading...
-
Application of Machine Learning Method in Classification of Rock Types in Open Pit Mines
- Publisher: European Association of Geoscientists & Engineers
- Source: Conference Proceedings, 8th Congress of the Balkan Geophysical Society, Oct 2015, Volume 2015, p.1 - 5
Abstract
Summary
Support vector (SV) method for classification originates from supervised machine learning methods. Although theoretically developed in the 70-ties of the 20-th century it was significantly improved in theory and practically implemented in the late 90-ties. Originally intended and elaborated as two class separation procedure it was latter transformed in robust multiclass classification technique. In this research the SV based technique for classification has been used for discrimination of rock types found in and around the open pit mines of Asarel-Medet mining complex located in the Srednogorie copper-porphyry mining region. The data used for the experiments are from the multispectral instruments TM/ETM+ onboard Landsat satellites from the same season of two different years. For ground trutning polygons having nomenclature under CORINE EU project were taken and adapted to the needs of this research. The results after classifying the area under study confirmed that the method selected is robust and offers good alternative to other approaches used for this task. In conclusions it is mentioned that for improvement of the outputs better spatial resolution is essential, but also more and narrower spectral bands would be offer an advantage.
© EAGE Publications BV