Discrete curvatures combined with machine learning for automated extraction of impact craters on 3D topographic meshes

Discrete curvatures combined with machine learning for automated extraction of impact craters on 3D topographic meshes

Abstract

We propose a new approach, based on the detection of crater rim, which form a characteristic round shape. The proposed approach contains two main steps: 1) each vertex is labelled with the values of the mean curvature and minimal curvatures; 2) this curvature map is injected into a Neural Network (NN) to automatically process the region of interest. As a NN approach, it requires a training set of manually detected craters to estimate the optimal weights of the NN. Once trained, the NN can be applied onto the regions of interest for automatically extracting all the craters.

Authors

  • Nicole Christoff
  • Laurent Jorda
  • Sophie Viseur
  • Sylvain Bouley
  • Agata Manolova
  • Jean-Luc Mari

Venue

EGU General Assembly Conference Abstracts, 2017.

Links

https://www.researchgate.net/publication/318099423_Discrete_curvatures_combined_with_machine_learning_for_automated_extraction_of_impact_craters_on_3D_topographic_meshes

Categories

, , , ,