正在加载图片...
D0L:10.13374.issn1001-053x.2013.09.016 第35卷第9期 北京科技大学学报 Vol.35 No.9 2013年9月 Journal of University of Science and Technology Beijing Sep.2013 基于Contourlet变换的连铸坯表面缺陷识别 徐科☒,艾永好,周鹏,杨朝霖 北京科技大学高效轧制因家工程研究中心,北京100083 ☒通信作者,E-mai让:xuke@ustb.edu.cm 摘要根据连铸坯表面图像的特点,提出了一种基于Contourlet变换的连铸坯表面缺陷识别方法.通过Contourlet变 换将样本图像分解成不同尺度和方向的子带,提取子带的Contourlet系数特征,并结合样本图像的纹理特征,得到一个 高维的特征向量.利用监督核保局投影算法对高维特征向量进行降维,将降维后的低维特征向量输入支持向量机,对连 铸坯表面图像进行分类识别.对现场采集到的裂纹、氧化铁皮、光照不均和渣痕四类样本图像进行实验,本文提出的识 别方法对样本图像的识别率可达94.35%,优于基于Gbor小波的识别方法. 关键词连铸坯;表面缺陷:模式识别:特征提取 分类号TP391.4 Recognition of surface defects in continuous casting slabs based on Contourlet transform XU Ke,AI Yong-hao,ZHOU Peng,YANG Chao-lin National Engineering Research Center for Advanced Rolling Technology,University of Science and Technology Beijing,Beijing 100083,China Corresponding author,E-mail:xuke@ustb.edu.cn ABSTRACT A new recognition method of surface defects based on Contourlet transform was proposed according to the characteristics of continuous casting slabs.Sample images were decomposed into multiple subbands with different scales and directions by Contourlet transform.The Contourlet coefficients of subbands and the textural features of sample images were combined into a high-dimensional feature vector.Supervised kernel locality preserving projection(SKLPP) was applied to the high-dimensional feature vector for dimension reduction,which resulted in a low-dimensional feature vector.The resulted feature vector was inputted to a support vector machine (SVM)for recognition of surface defects. The method was tested with sample images from an industrial production line,including cracks,scales,non-uniform illumination,and slags.The test results show that the recognition rate of these sample images is 94.35%,which is better than that by Gabor wavelet. KEY WORDS continuous casting slabs;surface defects;pattern recognition;feature extraction 目前,基于机器视觉的表面检测方法由于具有分类.基于机器视觉的表面检测方法主要通过表面 非接触、响应快等特点,已经广泛应用于多种工业图像中存在的信息来判断是否存在着缺陷以及缺陷 产品表面缺陷的检测与识别中-习.这种方法的特的类型,因此应尽可能全面地获取图像的边缘或纹 点是通过采集处于高强度光源照射下的工业产品表理等有效信息.针对不同的工业产品,所采用的具 面图像,借助图像处理和模式识别算法,对表面图 体识别方法也应有所不同,必须考虑该类产品的表 像进行分析,检测表面存在的缺陷,并对缺陷进行 面图像本身具有的特点. 收稿日期:2012-08-05 基金项目:教育部新世纪优秀人才支持计划资助项目(NCET-08-0726):北京市科技新星资助项目(2007B027):中央高校基本科研 业务费专项资金资助项目(FRF-TP09-027B)第 35 卷 第 9 期 北 京 科 技 大 学 学 报 Vol. 35 No. 9 2013 年 9 月 Journal of University of Science and Technology Beijing Sep. 2013 基于 Contourlet 变换的连铸坯表面缺陷识别 徐 科 ,艾永好,周 鹏,杨朝霖 北京科技大学高效轧制国家工程研究中心,北京 100083 通信作者,E-mail: xuke@ustb.edu.cn 摘 要 根据连铸坯表面图像的特点,提出了一种基于 Contourlet 变换的连铸坯表面缺陷识别方法. 通过 Contourlet 变 换将样本图像分解成不同尺度和方向的子带,提取子带的 Contourlet 系数特征,并结合样本图像的纹理特征,得到一个 高维的特征向量. 利用监督核保局投影算法对高维特征向量进行降维,将降维后的低维特征向量输入支持向量机,对连 铸坯表面图像进行分类识别. 对现场采集到的裂纹、氧化铁皮、光照不均和渣痕四类样本图像进行实验,本文提出的识 别方法对样本图像的识别率可达 94.35%,优于基于 Gabor 小波的识别方法. 关键词 连铸坯;表面缺陷;模式识别;特征提取 分类号 TP391.4 Recognition of surface defects in continuous casting slabs based on Contourlet transform XU Ke , AI Yong-hao, ZHOU Peng, YANG Chao-lin National Engineering Research Center for Advanced Rolling Technology, University of Science and Technology Beijing, Beijing 100083, China Corresponding author, E-mail: xuke@ustb.edu.cn ABSTRACT A new recognition method of surface defects based on Contourlet transform was proposed according to the characteristics of continuous casting slabs. Sample images were decomposed into multiple subbands with different scales and directions by Contourlet transform. The Contourlet coefficients of subbands and the textural features of sample images were combined into a high-dimensional feature vector. Supervised kernel locality preserving projection (SKLPP) was applied to the high-dimensional feature vector for dimension reduction, which resulted in a low-dimensional feature vector. The resulted feature vector was inputted to a support vector machine (SVM) for recognition of surface defects. The method was tested with sample images from an industrial production line, including cracks, scales, non-uniform illumination, and slags. The test results show that the recognition rate of these sample images is 94.35%, which is better than that by Gabor wavelet. KEY WORDS continuous casting slabs; surface defects; pattern recognition; feature extraction 目前,基于机器视觉的表面检测方法由于具有 非接触、响应快等特点,已经广泛应用于多种工业 产品表面缺陷的检测与识别中[1−2] . 这种方法的特 点是通过采集处于高强度光源照射下的工业产品表 面图像,借助图像处理和模式识别算法,对表面图 像进行分析,检测表面存在的缺陷,并对缺陷进行 分类. 基于机器视觉的表面检测方法主要通过表面 图像中存在的信息来判断是否存在着缺陷以及缺陷 的类型,因此应尽可能全面地获取图像的边缘或纹 理等有效信息. 针对不同的工业产品,所采用的具 体识别方法也应有所不同,必须考虑该类产品的表 面图像本身具有的特点. 收稿日期:2012–08–05 基金项目:教育部新世纪优秀人才支持计划资助项目 (NCET-08-0726);北京市科技新星资助项目 (2007B027);中央高校基本科研 业务费专项资金资助项目 (FRF-TP-09-027B) DOI:10.13374/j.issn1001-053x.2013.09.016
向下翻页>>
©2008-现在 cucdc.com 高等教育资讯网 版权所有