3D Reconstruction application in Quarrying industry

July 20, 2021

For the past 2 decades, the desire to produce 3D models of the world from 2D images has been researched in order to recover 3D point clouds based on Structure from motion and Multiview stereo methods. Meanwhile, the exploitation of minerals through quarrying, has been a lucrative market and a major contribution to the infrastructures of the cities since before the 18th century. The aim of this study is to apply 3D reconstruction approaches to limestone quarries. Building 3D model of quarry limestones from 2D images to be implemented in the quarry industry in favor of improving and modernizing the cleaving process through computer vision. The 2D images were acquired by a stereo camera pair, in a similar uncontrolled environment of an open-mine pit. In order to create fitting 2D images for structure from motion applications, it was necessary to explore different preprocessing approaches. Preprocessing methods were analyzed, and a gamma correction with a histogram equalization was deemed to be the most successful approach. Once the images were adjusted in terms of brightness and contrast, different SFM and MVS pipelines were tested to conclude which is the best available software application and its correspondent algorithm sequence. Although, the open-source software, Meshroom, was determined to be the best 3D reconstruction system, Agisoft Metashape can be more appropriate for industry applications due to its community support and convenient renewability. Nevertheless, Meshroom allows to study different matching algorithms, with known and unknown intrinsic and extrinsic parameters to rule the most effective approach.

Authors in the community:

• 

  Á

  Ágata Filipa Lopes de Barros

  ist426372

Supervisors of this institution:

• 

  J

  Jorge Manuel Mateus Martins

  ist14135

mechanical-engineering - Mechanical engineering

por - Portuguese

Instituto Superior Técnico