Strip Surface Defect Image Classification Based on Double-limited and Supervised-connect Isomap Algorithm
-
摘要: 根据带钢表面缺陷图像具有复杂纹理结构、包含大量干扰信息、具备高维非线性几何结构等特点,本文提出基于监督双限制连接Isomap方法的带钢表面缺陷图像降维方法(dls-Isomap).该方法以Isomap降维方法为基础,对其邻域图的连接方式进行K邻域(K-nearest neighbor,KNN)和ε-半径两个方面的限制性连接,并使用数据类别作为监督对类间邻域点进行扩展连接.针对多类Roll-swiss数据实验表明,dls-Isomap降维方法不仅能够在低维空间中完整嵌入所有数据点,而且能保持数据各类内和类间的几何结构,以及解决Isomap算法存在的“短路边”问题;针对带钢表面缺陷图像分类实验表明,基于dls-Isomap的新分类方法适合含水、油渍等干扰较多的带钢表面缺陷的分类任务,其中冷轧带钢5类缺陷识别率可以达78%.含水渍的热轧带钢缺陷识别率可以达到93%,其中水渍干扰图像的识别率达到97.6%.Abstract: A double-limited and supervised-connect Isomap dimensionality reduction and classification method (dls-Isomap) is proposed in this paper to classify more accurately the stripe surface defect images with the typical characteristics of complex texture, much noise, and high-dimension non-linear geometry. Based on the dimensionality reduction technique from Isomap, the connection of neighborhood graph is limited by key parameters K-nearest neighbor (KNN) and ε-radius, and inter-class neighborhood points are connected extensionally with the supervision of class labels. According to multi-classes roll-swiss data experiments, all the points can be embedded in lower dimensions with the complete inter-class and intra-class geometric structure, and the "short circuit" in the Isomap can be solved by the dls-Isomap method. In addition, stripe surface defect images data experiments show that the proposed classification method is suitable for the classification of stripe surface defects including more water and oil, with a recognition rate of 78% for cold-roll strip images, and 93% for hot-roll strip images with water, among which the recognition rata of water defects is 97.6%.
-
Key words:
- Isomap /
- K-nearest neighbor (KNN) /
- ε-radius /
- supervised-connect /
- stripe surface defect
-
[1] Ren Hai-Peng, Ma Zhan-Feng. Strip steel surface defect recognition based on complex network characteristics. Acta Automatica Sinica, 2011, 37(11): 1407-1412 (任海鹏, 马展峰. 基于复杂网络特性的带钢表面缺陷识别. 自动化学报, 2011, 37(11): 1407-1412) [2] Zhu Xian-Qiang, Shao Zhen-Feng, Li De-Ren. A rotation-invariant texture retrieval algorithm based on parameterless statistical features. Geomatics and Information Science of Wuhan University, 2010, 35(11): 1279-1282 (朱先强, 邵振峰, 李德仁. 利用无参数统计特征进行旋转不变纹理图像渐进检索. 武汉大学学报(信息科学版), 2010, 35(11): 1279-1282) [3] Shao Zhen-Feng, Li De-Ren, Zhu Xian-Qiang. Adaptive target detection based on improved genetic algorithm in infrared images. Geomatics and Information Science of Wuhan University, 2011, 36(5): 535-539 (邵振峰, 李德仁, 朱先强. 利用改进的遗传算法对红外影像自适应目标进行检测. 武汉大学学报(信息科学版), 2011, 36(5): 535-539) [4] Song Qiang, Xu Ke, Xu Jin-Wu. Recognition of surface defects on medium and heavy plates based on structure spectrum. Journal of University of Science and Technology Beijing, 2007, 29(3): 342-345 (宋强, 徐科, 徐金梧. 基于结构谱的中厚板表面缺陷识别方法. 北京科技大学学报, 2007, 29(3): 342-345) [5] Wu Xiu-Yong, Xu Ke, Xu Jin-Wu. Automatic recognition method of surface defects based on Gabor wavelet and kernel locality preserving projections. Acta Automatica Sinica, 2010, 36(3): 438-441 (吴秀永, 徐科, 徐金梧. 基于Gabor小波和核保局投影算法的表面缺陷自动识别方法. 自动化学报, 2010, 36(3): 438-441) [6] Orsenigo C, Vercellis C. An effective double-bounded tree-connected isomap algorithm for microarray data classification. Pattern Recognition Letters, 2012, 33(1): 9-16 [7] Li Wen-Feng, Xu Ke, Yang Chao-Lin, Gao Yang, Zhou Peng. Classier design of online surface defect inspection system for plates. Iron and Steel, 2006, 41(4): 47-50 (李文峰, 徐科, 杨朝霖, 高阳, 周鹏. 中厚板表面缺陷在线检测系统的分类器设计. 钢铁, 2006, 41(4): 47-50) [8] Tenenbaum J B, De Silva V, Langford J C. A global geometric framework for nonlinear dimensionality reduction. Science, 2000, 290(5500): 2319-2323 [9] Roweis S T, Saul L K. Nonlinear dimensionality reduction by locally linear embedding. Science, 2000, 290(5500): 2323-2326 [10] Belkin M, Niyogi P. Laplacian eigenmaps for dimensionality reduction and data representation. Neural Computation, 2003, 15(6): 1373-1396 [11] Yan De-Qin, Liu Sheng-Lan, Li Yan-Yan. An embedding dimension reduction algorithm based on sparse analysis. Acta Automatica Sinica, 2011, 37(11): 1306-1312 (闫德勤, 刘胜蓝, 李燕燕. 一种基于稀疏嵌入分析的降维方法. 自动化学报, 2011, 37(11): 1306-1312) [12] Zhu Mei-Qiang, Cheng Yu-Hu, Li Ming, Wang Xue-Song, Feng Huan-Ting. A hybrid transfer algorithm for reinforcement learning based on spectral method. Acta Automatica Sinica, 2012, 38(11): 1765-1776 (朱美强, 程玉虎, 李明, 王雪松, 冯涣婷. 一类基于谱方法的强化学习混合迁移算法. 自动化学报, 2012, 38(11): 1765-1776) [13] Tao Jian-Wen, Wang Shi-Tong. Locality-preserved maximum information variance V-support vector machine. Acta Automatica Sinica, 2011, 38(1): 97-108 (陶剑文, 王士同. 局部保留最大信息差V-支持向量机. 自动化学报, 2012, 38(1): 97-108) [14] Vapnik V. The Nature of Statistical Learning Theory. New York: Springer-Verlag, 1995. 8-10 [15] Geng X, Zhan D C, Zhou Z H. Supervised nonlinear dimensionality reduction for visualization and classification. IEEE Transactions on Systems, Man, and Cybemetics Part B——Cybernetics, 2005, 35(6): 1098-1107 [16] Gan Sheng-Feng, Sun Lin. Application of multi-level image classification system in surface defect detection of cold-rolled silicon stripe. Metallurgical Industry Automation, 2009, 33(2): 63-65 (甘胜丰, 孙林. 多级图像分类系统在硅钢冷轧表面缺陷检测中的应用. 冶金自动化, 2009, 33(2): 63-65) [17] Cox T F, Cox M A A. Multidimensional Scaling. London: Chapman and Hall, 1994. 57-66 [18] Svetlana L, Milica J, Zorica D, Jelena P, Ljiljana S. The application of generalized regression neural network in the modeling and optimization of aspirin extended release tablets with Eudragit RS PO as matrix substance. Journal of Controlled Release, 2002, 82: 213-222
点击查看大图
计量
- 文章访问数: 1809
- HTML全文浏览量: 59
- PDF下载量: 1624
- 被引次数: 0