卷對卷式鋼帶表面瑕疵智慧光學檢測技術

Smart Optical Inspection Technology for Surface Defects of Strip Steel in Roll-to-Roll Processing

陳柏安 ； 劉定坤 ； 吳潔瑜 ； 王柏凱

《機械工業雜誌》 475期 (2022/10) Pp. 63-69

摘要

卷對卷式鋼帶在經過壓延、退火酸洗與分條加工程序後，鋼材表面可能產生刮痕與鑿點的瑕疵。過去多經由人員在鋼帶卷對卷式傳輸過程時，以肉眼觀察表面是否有瑕疵，但不僅容易造成人眼疲勞，而判斷的標準也常因環境因素或人員差異而產生不一致的狀況，且在操作的過程也增加可能發生工安意外的風險。為了達到即時監測品質的一致性，本文發展一套智慧光學檢測技術以取代傳統人力檢測的方式，以高角度與低角度兩種光源形式同時打光，並利用配置不同波長的光源與彩色線型相機光譜響應不重疊的特性，同時拍攝兩種光源特性以強化瑕疵特徵，最後利用卷積運算、影像比對與機器學習中的支援向量機（Support Vector Machine, SVM）進行分類以完成檢測。本技術在影像解析度30 μm、最大對應鋼卷寬度490 mm、最大輸送速度2 m/s的情況下，可實現即時全檢的高速智慧檢測。

關鍵字

鋼鐵表面瑕疵 ； 卷對卷 ； 智慧光學檢測

並列摘要

After rolling, annealing, acid cleaning and slitting, the strip steel may have scratches and chisel defects on the steel surface. In the past, the strip steel surface defects were visually inspected by human during the roll-to-roll transmission process. Eye fatigue causes judgment criteria inconsistency among different people; and the risk of industrial accidents increase during the operation. In order to achieve the consistency of judgment criteria, this work develops a set of automatic optical inspection technology to replace traditional visual inspection. The surface defect characteristics of strip steel are enhanced by using light illumination with two different angles and wavelengths. Finally, the detection is completed by convolution operation, image comparison, and support vector machine classification. With image resolution of 30 μm, high-speed and real-time full inspection can be realized, achieving maximum steel coil of 490 mm width and the maximum conveying speed of 2 m/s.

並列關鍵字

Steel surface defects ； Roll to roll ； Smart optical inspection

參考文獻

劉曉薇、劉定坤、藍于瀅、李漢文、陳中塘，“ 機器視覺應用於機械加工件表面線上檢測 ”, 2016 AOI 論壇與展覽。

R. Keys, “Cubic convolution interpolation for digital image processing,” in IEEE Transactions on Acoustics, Speech, and Signal Processing 29(6), 1153-1160, 1981.

V. Podlozhnyuk, “Image convolution with CUDA,” NVIDIA Corporation, 2007.

O. R. Vincent and O. Folorunso, “A descriptive algorithm for sobel image edge detection,” in Proceedings of Informing Science & IT Education Conference (InSITE 2009) 9, 12-15, 2009.

R. S. Montero and E. Bribiesca, “State of the art of compactness and circularity measures,” International Mathematical Forum 4(27), 1305-1335, 2009.