Digital Geometry in Pattern Recognition: Extracting Geometric Features with DGtal and Applications

Abstract

Estimating geometric parameters can play an important role in different applications of pattern recognition algorithms. The purpose of this tutorial is to present new efficient and easy to reproduce algorithms dedicated to the robust extraction of geometric features on digital objects. The tutorial will consist of a rapid overview of the theoretical description of recent estimators (like tangent, curvature or noise estimators) followed by the implementation details given through the DGtal library. To introduce this library, a quick look of the main features will be presented with preliminary exercises available online before the tutorial. To simplify interaction and demonstration like in teaching activities, the use of the DGtal library will be embedded in a Jupyter notebook allowing online demonstration from a single web browser. In the continuity and complement of this online notebook, we will propose to the audience to construct its own online demonstration by using the new automatic demonstration kernel recently proposed in the Image Processing On Line journal.

Jacques-Olivier Lachaud

Professor of Computer Science

My research interests include digital geometry, geometry processing, image analysis, variational models and discrete calculus.