基于全局-局部SVM的钢板表面缺陷分类
|
摘要
为了提高钢板表面缺陷分类识别,提出一种新的全局和局部支持向量机分类模型。首先,通过样本提取算法将整个训练样本划分为非边界样本和边界样本。非边界样本用于训练全局支持向量机,并获得两条全局决策边界。边界样本用于寻找测试样本的K-近邻样本,通过训练K-近邻样本获得相应的局部支持向量机。若测试样本位于全局决策边界线两侧,直接给出分类结果,否则,由局部支持向量机进行分类决策。最终,新的模型结合二叉树算法实现了4种钢板表面缺陷的分类问题。实验结果显示,全局和局部支持向量机模型有令人满意的综合性能。
In order to improve steel surface defect classification performance,a novel global and local support vector machines( GLSVMs) model is proposed in this paper. Firstly,the whole training samples are divided into boundary and non-boundary samples. Non-boundary samples are used to train global SVM and obtain two global decision boundaries. Boundary samples are used to search K-nearest neighbors of testing samples,and local SVM can be obtained by training K-nearest neighbor samples. If a testing sample was beside of global decision boundary,then its defect type can be directly determined,otherwise,the classification result is determined by local SVM. Finally,the novel model is combined with binary tree to realize four types of steel surface defects classification. The experimental results show that the GLSVMs model has satisfactory performance.
In order to improve steel surface defect classification performance,a novel global and local support vector machines( GLSVMs) model is proposed in this paper. Firstly,the whole training samples are divided into boundary and non-boundary samples. Non-boundary samples are used to train global SVM and obtain two global decision boundaries. Boundary samples are used to search K-nearest neighbors of testing samples,and local SVM can be obtained by training K-nearest neighbor samples. If a testing sample was beside of global decision boundary,then its defect type can be directly determined,otherwise,the classification result is determined by local SVM. Finally,the novel model is combined with binary tree to realize four types of steel surface defects classification. The experimental results show that the GLSVMs model has satisfactory performance.
引文
[1]Luo Q,He Y. A cost-effective and automatic surface defect inspection system for hot-rolled flat steel[J]. Robotics Computer-Integrated Manuf,2016,38:16-30.
[2]Gong R,Chu M,Wang A,et al. A fast detection method for region of defect on strip steel surface[J]. ISIJ Int,2015,55(1):207-212.
[3]Hu H,Li Y,Liu M,et al. Classification of defects in steel strip surface based on multiclass support vector machine[J]. Multimedia tools and applications,2014,69(1):199-216.
[4]Martins L A O,Padua F L C,Almeida P E M. Automatic detection of surface defects on rolled steel using computer vision and artificial neural networks[C]. Proceedings of 36thAnnual Conference on Ieee Industrial Electronics Society,Glendale,AZ,2010:1081-1086.
[5] Agarwal K,Shivpuri R,Zhu Y J,et al. Process knowledge based multi-class support vector classification(PK-MSVM)approach for surface defects in hot rolling[J]. Expert Systems with,2011,38(6):7251-7262.
[6]Chu M,Zhao J,Liu X,et al. Multi-class classification for steel surface defects based on machine learning with quantile hyper-spheres[J].Chemometr.Intell.Lab.Syst,2017,168:15-27.
[7]Suvdaa B,Ahn J,Ko J. Steel surface defects detection and classification using SIFT and voting strategy[J]. International Journal of Software Engineering and Its Applications,2012,6(2):161-166.
[8] Chu M,Gong R,Wang A. Strip steel surface defect classification method based on enhanced twin support vector machine[J]. ISIJ international,2014,54(1):119-124.
[9]Cortes C,Vapnik V. Support-vector networks[J]. Machine learning,1995,20(3):273-297.
[10]尹传环,牟少敏,田盛丰,等.局部支持向量机的研究进展[J].计算机科学,2012,39(1):170-174,189.
[11] Liu L,Wang A,Mo S,et al. Multi-class classification methods of cost-conscious ls-svm for fault diagnosis of blast furnace[J]. Journal of Iron and Steel Research,International,2011,18(10):17-23,33.
[2]Gong R,Chu M,Wang A,et al. A fast detection method for region of defect on strip steel surface[J]. ISIJ Int,2015,55(1):207-212.
[3]Hu H,Li Y,Liu M,et al. Classification of defects in steel strip surface based on multiclass support vector machine[J]. Multimedia tools and applications,2014,69(1):199-216.
[4]Martins L A O,Padua F L C,Almeida P E M. Automatic detection of surface defects on rolled steel using computer vision and artificial neural networks[C]. Proceedings of 36thAnnual Conference on Ieee Industrial Electronics Society,Glendale,AZ,2010:1081-1086.
[5] Agarwal K,Shivpuri R,Zhu Y J,et al. Process knowledge based multi-class support vector classification(PK-MSVM)approach for surface defects in hot rolling[J]. Expert Systems with,2011,38(6):7251-7262.
[6]Chu M,Zhao J,Liu X,et al. Multi-class classification for steel surface defects based on machine learning with quantile hyper-spheres[J].Chemometr.Intell.Lab.Syst,2017,168:15-27.
[7]Suvdaa B,Ahn J,Ko J. Steel surface defects detection and classification using SIFT and voting strategy[J]. International Journal of Software Engineering and Its Applications,2012,6(2):161-166.
[8] Chu M,Gong R,Wang A. Strip steel surface defect classification method based on enhanced twin support vector machine[J]. ISIJ international,2014,54(1):119-124.
[9]Cortes C,Vapnik V. Support-vector networks[J]. Machine learning,1995,20(3):273-297.
[10]尹传环,牟少敏,田盛丰,等.局部支持向量机的研究进展[J].计算机科学,2012,39(1):170-174,189.
[11] Liu L,Wang A,Mo S,et al. Multi-class classification methods of cost-conscious ls-svm for fault diagnosis of blast furnace[J]. Journal of Iron and Steel Research,International,2011,18(10):17-23,33.