Surface defect recognition for moderately thick plates based on a SIFT operator
ZHOU Peng1, XU Ke2, YANG Chaolin1
1. Institute of Engineering Technology, University of Science & Technology Beijing, Beijing 100083, China; 2. Collaborative Innovation Center of Iron and Steel Technology, University of Science and Technology Beijing, Beijing 100083, China
Abstract：Moderately thick plates have defects such as indentations, roll marks and scratches from the production processes. Since serious defects can negatively impact the next rolling process, operators must identify serious defects containing iron oxide defects on the surface to improve the quality of iron and steel products. A scale invariant feature transform (SIFT) operator was used to extract feature vectors that are scale rotation invariant. A Euclidean distance similarity measure is used for image matching to identify the surface defects of moderately thick plates. Many tests were then run to identify the proper values of each parameter. The SIFT algorithm then had a 95% surface defect recognition rate and was especially effective for continuous defects. Thus, this SIFT method which is unaffected by the illumination and is rotation and affine invariant gives excellent recognition of iron oxide defects.
周鹏, 徐科, 杨朝霖. 基于SIFT的中厚板表面缺陷识别方法[J]. 清华大学学报（自然科学版）, 2018, 58(10): 881-887.
ZHOU Peng, XU Ke, YANG Chaolin. Surface defect recognition for moderately thick plates based on a SIFT operator. Journal of Tsinghua University(Science and Technology), 2018, 58(10): 881-887.
[1] 吴平川, 路同浚, 王炎. 机器视觉与钢板表面缺陷的无损检测[J]. 无损检测, 2000, 22(1):13-16, 47. WU P C, LU T J, WANG Y. Machine-vision technology and nondestructive detection of the surface defects in strip steel[J]. Nondestructive Testing, 2000, 22(1):13-16, 47. (in Chinese) [2] 徐科, 徐金梧. 基于图像处理的冷轧带钢表面缺陷在线检测技术[J]. 钢铁, 2002, 37(12):61-64. XU K, XU J W. On-line inspection of surface defects of cold rolled strips based on image processing[J]. Iron & Steel, 2002, 37(12):61-64. (in Chinese) [3] LOWE D G. Object recognition from local scale-invariant features[C]//Proceedings of the 17th IEEE International Conference on Computer Vision. Kerkyra, Greece:IEEE, 1999:1150-1157. [4] LINDEBERG T. Feature detection with automatic scale selection[J]. International Journal of Computer Vision, 1998, 30(2):79-116. [5] WANG H X, YANG K J, GAO F, et al. Normalization methods of SIFT vector for object recognition[C]//Proceedings of the 10th International Symposium on Distributed Computing and Applications to Business, Engineering and Science. Wuxi, China:IEEE, 2011:175-178. [6] ZHAO W L, NGO C W. Flip-invariant SIFT for copy and object detection[J]. IEEE Transactions on Image Processing, 2013, 22(3):980-991. [7] BELCHER C, DU Y Z. Region-based SIFT approach to iris recognition[J]. Optics and Lasers in Engineering, 2009, 47(1):139-147. [8] YUE H J, SUN X Y, WU F, et al. SIFT-based image compression[C]//Proceedings of 2012 IEEE International Conference on Multimedia and Expo. Melbourne, VIC, Australia:IEEE, 2012:473-478. [9] HSU C Y, LU C S, PEI S C. Image feature extraction in encrypted domain with privacy-preserving SIFT[J]. IEEE Transactions on Image Processing, 2012, 21(11):4593-4607. [10] OBESO F, GONZALEZ J A, BROWN A. Intelligent on-line surface inspection on a skinpass mill[J]. Iron and Steel Engineer, 1997(10):29-35. [11] LOWE D G. Distinctive image features from scale-invariant keypoints[J]. International Journal of Computer Vision, 2004, 60(2):91-110.
2018, Vol. 58 
