Title :
On-line defect detection algorithms for surface inspection of hot rolled strips
Author :
Ke, Xu ; Chaolin, Yang
Author_Institution :
Nat. Eng. Res. Center of Adv. Rolling, Univ. of Sci. & Technol. Beijing, Beijing, China
Abstract :
As there are a lot of scales, water drops and color differences on the surface of hot rolled strips, traditional algorithms of defect detection is not applicable for on-line surface inspection of hot rolled strips. New algorithms are developed, including three steps, which are image segmentation, four types of detection, and defect classification. Image segmentation is in real-time thread, while four types of detection are in just-in-time thread and defect classification is in idle-time thread. Three steps are implemented in the parallel mode for on-line inspection. The algorithms are applied to a newly developed surface inspection system of hot rolled strips, and time costs of the algorithms are bellow needs of on-line inspection.
Keywords :
hot rolling; image segmentation; inspection; production engineering computing; strips; defect classification; hot rolled strip; idle-time thread; image segmentation; just-in-time thread; online defect detection; surface inspection; Charge coupled devices; Charge-coupled image sensors; Costs; Detection algorithms; Fault detection; Image segmentation; Inspection; Strips; Surface emitting lasers; Yarn; defect detetion; hot rolled strips; surface inspection;
Conference_Titel :
Mechanic Automation and Control Engineering (MACE), 2010 International Conference on
Conference_Location :
Wuhan
Print_ISBN :
978-1-4244-7737-1
DOI :
10.1109/MACE.2010.5535586