Research on the segmentation method of surface defects of galvanised sheet under complex texture background

Abstract

Zinc-coated steel sheets, a kind of high-quality sheet, is widely used in all sorts of manufacturing industry. The fault on its surface is one of the factors which exert bad effects on its quality. Due to its complex texture, the experts first put forward to the defect segmentation methods of Discrete Fourier Spectrum Residuals and Gabor Filtering Image Fusion. Then, they analysed the grey form of the faults and the deficiencies of the two methods and made some improvements on them using Morphological black hat operation and eventually achieved the goal of effective defect segmentation. According to some results of experiments, the improved method is of high accuracy in the defect segmentation and to some extent can serve as the reference method to examine the faults of the zinc-coated sheets.

Keywords

• steel
• image texture
• image segmentation
• image fusion
• automatic optical inspection
• gabor filters
• feature extraction
• zinc
• galvanising
• sheet materials
• zinc-coated sheets
• effective defect segmentation
• morphological black hat operation
• grey form
• gabor filtering image fusion
• discrete fourier spectrum residuals
• defect segmentation methods
• bad effects
• manufacturing industry
• sorts
• high-quality sheet
• zinc-coated steel sheets
• complex texture background
• galvanised sheet
• surface defects
• segmentation method