P.A. Hölzl, D.C.H. Schleicher, and B.G. Zagar (Austria)
pattern recognition, machine vision, high dynamic range imaging
Due to the ever increasing quality levels in the steel in dustry quality control via machine vision systems gets ever more important. In this paper a robust algorithm to detect the coils outline, straps and packaging under natural illumination is presented. In contrast to machine vision applications operating under well defined lighting conditions, in steel industry such illumination can not be applied without enormous costs. Coping with these adverse lighting conditions dramatically in creases the algorithmic burden on the image processing task. We show that generating and using high dynamic range images and a priori knowledge about the scene, enables a satisfying object recognition protocoll.
Important Links:
Go Back
