Community Research and Development Information Service - CORDIS


The general objective of this project was to improve surface inspection of moving steel products by carefully looking at sensors and lighting, image processing and segmentation by texture analysis methods, and building of adaptive defect classifiers with knowledge-based techniques. In order to be as complete as possible, different kinds of steel products and processes were selected. The sensor configurations and lighting conditions were studied. Defect samples were collected (databases) and analysed by human experts. The databases were then used for image processing and classification tuning. Apart from the performance gain obtained for each product and process, the overall result was a common methodology, which could be a kind of guideline for the building and tuning of automatic surface inspection systems on flat steel products. This methodology was applied successfully within the project on the inspection of cold-rolled strip after continuous annealing at the Sollac Florange plant.

Additional information

Authors: TOURSCHER G, IRSID, Maizieres-les-Metz (FR);BLANCHARD D, IRSID, Maizieres-les-Metz (FR);ODET C, INS Creatis, Villeurbanne (FR);SEGALINI S, Soliac, Montataire (FR);ARBIOL M, Soliac, Montataire (FR);CALEB P, UWE-BTC, Bristol (GB): VDEh/BFI, Dusseldorf (DE);MENNICKEN H,
Bibliographic Reference: EUR 20082 EN (2002), pp.108. Euro: 16.00
Availability: EUR-OP reference: KI-NA-20082-EN-S. Available from EUR-OP sales agents. URL:
ISBN: ISBN: 92-894-2198-3
Record Number: 200214435 / Last updated on: 2002-02-20
Original language: en
Available languages: en