AUTOMATIC PETROGRAPHIC INSPECTION BY USING IMAGE PROCESSING AND MACHINE LEARNING

Katharina Anding, Daniel Garten, André Göpfert, Matthias Rückwardt, Edgar Reetz, Gerhard Linß

Abstract

In this paper we present a method for an automatic inspection of mineral aggregates. Certain components of aggregates can negatively affect the mechanical strength of produced concrete or asphalt as well as cause the destructive alkali silica reaction. The automatic recognition of such aggregates could be successfully implemented by means of image processing and machine learning algorithms. We achieved a total recognition rate of 88% for this very complex recognition problem. Therefore every step in the pattern recognition process has been successfully optimized. In conjunction with the object singularization we achieved a fully automatic testing instrument for petrographic inspection.

Keywords

aggregate, petrographic inspection, SVM, machine learning

Download

IMEKO-WC-2012-TC2-O4.pdf

DOI

-

IMEKO TC

TC2 - Photonics

Back to the proceedings