On May 7th at 12:00, Room B-221 (ETSIT)



News and Events 

Automatic mineral recognition from hyperspectral images of cores using a deep learning framework

On May 7th at 12:00, Room B-221.

New minerals have been discovered recently, such as the rare-earth elements, which awaken great interest in several companies which compete to find new deposits, especially in parts of the world less exploited, because of their application to the development of electronic components and devices: for example, mobile phones, chips, tablets or batteries. Several research lines propose solutions to automatize mineral recognition, considering that this process is usually performed manually, despite being crucial in the value chain of mining companies. Together with technological advances about recently developed sensors and innovative processing capabilities, automatic mineral recognition would make explorations more efficient and sustainable and would reduce considerable costs.

In this line, a deep learning framework for automatic mineral recognition from hyperspectral images of scanned cores was presented in this talk. It has been developing for the European public Project Innolog from EIT Raw Materials to improve downhole geophysical logging tools for real-time identification, evaluation and quantification of mineral deposits and raw materials in the subsurface. Fully convolutional neural networks were explained as one of the most widely used deep learning techniques for their great capability to comprehensive complex scene understanding. Finally, experimental results of mineral recognition obtained with databases of hyperspectral images of scanned cores for different infrared wavelength ranges were analysed.

Andrés Bell received the Bachelor of Engineering in Telecommunication Technologies and Services and the Master in Telecommunication Engineering from the Universidad Politécnica de Madrid (UPM), Madrid, Spain, in 2015 and in 2017 respectively. He is currently pursuing the Ph.D degree at the same University.

Since 2014, he has been a member of the Grupo de Tratamiento de Imágenes (Image Processing Group) at the UPM. His current research interests include computer vision, machine learning, deep learning, video and image analysis and processing.