摘要
提出了一种基于小波变换和灰度共生矩阵的带钢表面缺陷识别方法。采用小波变换分解缺陷图像并提取其低频子带信息。通过在低频子带上构造0°、45°、90°和135°四个方向的灰度共生矩阵,分别计算角二阶矩、熵、对比度和逆差矩4个特征值,共获得16个特征值,并将其输入支持向量机,完成对6类共1800张带钢表面缺陷图像的识别,总体识别精度大于96%。实验结果表明,小波变换与灰度共生矩阵结合能有效描述带钢表面缺陷纹理特征,具有较好的识别效果。
A method for recognition of surface defects on strip based on wavelet transform and gray level co-occurrence matrix(GLCM)are proposed.Wavelet transform is used to decompose defect image and extract their low frequency sub-band.By constructing gray level co-occurrence matrices in four directions of 0°,45°,90°and 135°on low frequency sub-band,four eigenvalues of angular second moment,contrast,entropy and inverse difference moment are calculated respectively,and obtain 16 eigenvalues,which are input into support vector machine(SVM),complete the recognition of surface defects on strip of 1800 images in six categories.The overall recognition accuracy is more than 96%.The experimental results show that the combination of wavelet transform and gray level co-occurrence matrix can effectively describe the texture characteristics of surface defects on strip and has a good recognition effect.
作者
单东日
童灿
乃学尚
高立营
王玉伟
SHAN Dongri;TONG Can;NAI Xueshang;GAO Liying;WANG Yuwei(School of Electronic and Information Engineering,Qilu University of Technology(Shandong Academy of Sciences),Jinan 250353,CHN;School of Mechanical and Automotive Engineering,Qilu University of Technology(Shandong Academy of Sciences),Jinan 250353,CHN;School of Information and Control Engineering,Shandong Vocational College of Foreign Affairs Translation,Weihai 264504,CHN)
出处
《制造技术与机床》
北大核心
2020年第2期120-123,共4页
Manufacturing Technology & Machine Tool
基金
山东省自主创新及成果转化专项(2015ZDXX0101G06)
山东省高等学校科技发展计划(J18KA032)
