Faulty Scarfing Slab Detection Using Machine Vision

Abstract

In this study, we developed a detection algorithm using machine vision for faulty scarfing on slabs. Scarfing is a process for burning out defective areas on the surface of semi-finished steel products so that the product is suitable for subsequent rolling or forging. In the case of rolling, a poorly scarfed slab can cause serious defects. Therefore, an automatic inspection system for scarfed slabs should be developed. In the image of faultily scarfed slab, discontinuous bright bands and borderlines are observed. To detect the bright bands, we examined the real part of a Gabor filter. However, because the size of the bright bands is not fixed, the parameters of the real part of the Gabor filter were determined adaptively using the imaginary part of the Gabor filter. The performance of the proposed algorithm was tested with 2292 images of scarfed slabs of which 49 of the images were of faulty scarfed slabs. The proposed algorithm was able to detect defects with 95.9% accurate and the false alarm was 1.6%.