A novel nonuniformity-correction-based defect detection method for the aluminum strip surface inspection system

作者：Luo, Xinbin; Fu, Shan; Huang, Xiuqin

来源：Journal of Computational and Theoretical Nanoscience, 2016, 13(1): 500-509.

DOI：10.1166/jctn.2016.4832

• Aluminum surface
• Defect detection
• Defect detection algorithm
• Defect detection method
• Gaussian model
• Nonuniformity correction
• Nonuniformity correction algorithm
• Statistical information

Traditional metal strip defect detection algorithms focus mainly on improving the detection algorithm itself, considering little the impact of the nonuniformity, which affects greatly the results of the defect detection for the metal strip surface. In this paper, we propose a novel nonuniformity-correctionbased defect detection method for a high-speed aluminum strip surface inspection system, by focusing on a new nonuniformity correction technique. A real-time image correction algorithm is designed to obtain a homogeneous background image, and an adaptive dual threshold segmentation algorithm is proposed to segment the defect target in the corrected image. The nonuniformity correction algorithm is a novel scene-based correction method that updates correction parameters through statistical information of the image sequence. Hence, it is able to remove not only the nonuniformity caused by the fixed pattern noise, but also that caused by the external environment. The nonuniformity correction is the foundation of the whole defect detection approach, which improves the contrast between the defect and the background. Experimental results show that the nonuniformitycorrection-based detection method is simple and efficient for metal strip surface defect detection, achieving real-time and highly accurate defect detection.

• 出版日期2016
• 单位上海交通大学