基于最大熵与密度聚类相融合的毛羽检测
|
摘要
为能够更加精确地计算出纱线毛羽的根数及毛羽长度,基于最大熵与密度聚类相融合对纱线毛羽的长度及根数进行检测。该方法首先利用双边滤波对采集到的纱线图像进行预处理,滤除图像中的噪声,同时增强纱线毛羽特征;然后利用最大熵对预处理后的纱线图像进行阈值分割,去除条干提取毛羽,并对毛羽进行细化;最后利用密度聚类算法(DBSCAN聚类)对细化后的毛羽进行分类统计,根据所分类的个数以及每类所含像素点的个数计算出毛羽的根数及长度。将实验结果与目测法和基准线法进行比较,结果表明,该方法与目测方法检测的结果非常接近,结果比基准线法更加精确,检测结果准确、有效。
In order to calculate the number of yarn hairiness and hairiness length more accurately,a method based on maximum entropy and density clustering was proposed to detect the yarn hairiness length and root number. The yarn image was preprocessed by bilateral filtering to filter out the noise in the image and enhance the yarn hairiness characteristics. Then the maximum entropy was adopted to segment the preprocessed yarn image and remove the yarn. The hairiness were extracted and refined. Finally,the density clustering algorithm( DBSCAN clustering) was applied to classify the number of hairiness. In addition,the root and length of hairiness according to the number of classified hairiness and the number of pixels in each class were calculated. Compared with the visual method and the datum line method,the experimental results demonstrate that the proposed method is very close to the visual method and more accurate than the datum line method. Furthermore,it is shown that the proposed method is accurate and effective.
In order to calculate the number of yarn hairiness and hairiness length more accurately,a method based on maximum entropy and density clustering was proposed to detect the yarn hairiness length and root number. The yarn image was preprocessed by bilateral filtering to filter out the noise in the image and enhance the yarn hairiness characteristics. Then the maximum entropy was adopted to segment the preprocessed yarn image and remove the yarn. The hairiness were extracted and refined. Finally,the density clustering algorithm( DBSCAN clustering) was applied to classify the number of hairiness. In addition,the root and length of hairiness according to the number of classified hairiness and the number of pixels in each class were calculated. Compared with the visual method and the datum line method,the experimental results demonstrate that the proposed method is very close to the visual method and more accurate than the datum line method. Furthermore,it is shown that the proposed method is accurate and effective.
引文
[1]杨红英,朱苏康.纱线毛羽[J].纺织学报,2000,21(6):11-14.YANG Hongying,ZHU Sukang.Yarn hairiness[J].Journal of Textile Research,2000,21(6):11-14.
[2]肖国兰.浅析纱线毛羽的成因及预防措施[J].上海纺织科技,2014,42(1):42-43,52.XIAO Guolan.Causes and pre-prevention measures of yarn hairiness[J].Shanghai Textile Science&Technology,2014,42(1):42-43,52.
[3]JING J,HUANG M,LI P,et al.Automatic measurement of yarn hairiness based on the improved MRMRF segmentation algorithm[J].Journal of The Textile Institute,2018,109(6):740-749.
[4]孙银银,潘如如,高卫东.基于数字图像处理的纱线毛羽检测[J].纺织学报,2013,34(6):102-106.SUN Yinyin,PAN Ruru,GAO Weidong.Detection of yarn hairiness based on digital image processing[J].Journal of Textile Research,2013,34(6):102-106.
[5]FABIJA N'SKA A,JACKOWSKA-STRUMIO L.Image processing and analysis algorithms for yarn hairiness determination[J].Machine Vision and Applications,2012,23(3):527-540.
[6]郭海涛,田坦,王连玉,等.利用二维属性直方图的最大熵的图像分割方法[J].光学学报,2006(4):506-509.GUO Haitao,TIAN Tan,WANG Lianyu,et al.Image segmentation method based on maximum entropy of twodimensional attribute histogram[J].Acta Optica Sinica,2006(4):506-509.
[7]伦向敏,侯一民.运用迭代最大熵算法选取最佳图像分割阈值[J].计算机工程与设计,2015,36(5):1265-1268,1289.LUN Xiangmin,HOU Yimin.Using the iterative maximum entropy algorithm to select the optimal image segmentation threshold[J].Computer Engineering and Design,2015,36(5):1265-1268,1289.
[8]HILDITCH C J.Comparison of thinning algorithms on a parallel processor[J].Image and Vision Computing,1983,1(3):115-132.
[9]李宗林.基于DBSCAN的自适应聚类算法研究[D].长沙:长沙理工大学,2015:14-23.LI Zonglin.Research on adaptive clustering algorithm based on DBSCAN[D].Changsha:Changsha University of Science and Technology,2015:14-23.
[10]周培培,丁庆海,罗海波,等.基于DBSCAN聚类算法的异常轨迹检测[J].红外与激光工程,2017,46(5):238-245.ZHOU Peipei,DING Qinghai,LUO Haibo,et al.Abnormal trajectory detection based on DBSCANclustering algorithm[J].Infrared and Laser Engineering,2017,46(5):238-245.
[11]孙银银,张宁,吴洋,等.纱线毛羽骨架及长度的跟踪测量[J].纺织学报,2017,38(8):32-38.SUN Yinyin,ZHANG Ning,WU Yang,et al.Tracking measurement of yarn hairiness skeleton and length[J].Journal of Textile Research,2017,38(8):32-38.
[12]苏继伟.简论纱线毛羽的测试方法[J].上海纺织科技,2004,32(6):60-62.SU Jiwei.A brief discussion on test method of yarn hairiness[J].Shanghai Textile Science Technology,2004,32(6):60-62.
[13]张继蕾.基于图像处理技术的纱线毛羽检测应用研究[D].石家庄:河北科技大学,2011:2.ZHANG Jilei.Research on yarn hairiness detection based on image processing technology[D].Shijiazhuang:Hebei University of Science and Technology,2011:2.
[2]肖国兰.浅析纱线毛羽的成因及预防措施[J].上海纺织科技,2014,42(1):42-43,52.XIAO Guolan.Causes and pre-prevention measures of yarn hairiness[J].Shanghai Textile Science&Technology,2014,42(1):42-43,52.
[3]JING J,HUANG M,LI P,et al.Automatic measurement of yarn hairiness based on the improved MRMRF segmentation algorithm[J].Journal of The Textile Institute,2018,109(6):740-749.
[4]孙银银,潘如如,高卫东.基于数字图像处理的纱线毛羽检测[J].纺织学报,2013,34(6):102-106.SUN Yinyin,PAN Ruru,GAO Weidong.Detection of yarn hairiness based on digital image processing[J].Journal of Textile Research,2013,34(6):102-106.
[5]FABIJA N'SKA A,JACKOWSKA-STRUMIO L.Image processing and analysis algorithms for yarn hairiness determination[J].Machine Vision and Applications,2012,23(3):527-540.
[6]郭海涛,田坦,王连玉,等.利用二维属性直方图的最大熵的图像分割方法[J].光学学报,2006(4):506-509.GUO Haitao,TIAN Tan,WANG Lianyu,et al.Image segmentation method based on maximum entropy of twodimensional attribute histogram[J].Acta Optica Sinica,2006(4):506-509.
[7]伦向敏,侯一民.运用迭代最大熵算法选取最佳图像分割阈值[J].计算机工程与设计,2015,36(5):1265-1268,1289.LUN Xiangmin,HOU Yimin.Using the iterative maximum entropy algorithm to select the optimal image segmentation threshold[J].Computer Engineering and Design,2015,36(5):1265-1268,1289.
[8]HILDITCH C J.Comparison of thinning algorithms on a parallel processor[J].Image and Vision Computing,1983,1(3):115-132.
[9]李宗林.基于DBSCAN的自适应聚类算法研究[D].长沙:长沙理工大学,2015:14-23.LI Zonglin.Research on adaptive clustering algorithm based on DBSCAN[D].Changsha:Changsha University of Science and Technology,2015:14-23.
[10]周培培,丁庆海,罗海波,等.基于DBSCAN聚类算法的异常轨迹检测[J].红外与激光工程,2017,46(5):238-245.ZHOU Peipei,DING Qinghai,LUO Haibo,et al.Abnormal trajectory detection based on DBSCANclustering algorithm[J].Infrared and Laser Engineering,2017,46(5):238-245.
[11]孙银银,张宁,吴洋,等.纱线毛羽骨架及长度的跟踪测量[J].纺织学报,2017,38(8):32-38.SUN Yinyin,ZHANG Ning,WU Yang,et al.Tracking measurement of yarn hairiness skeleton and length[J].Journal of Textile Research,2017,38(8):32-38.
[12]苏继伟.简论纱线毛羽的测试方法[J].上海纺织科技,2004,32(6):60-62.SU Jiwei.A brief discussion on test method of yarn hairiness[J].Shanghai Textile Science Technology,2004,32(6):60-62.
[13]张继蕾.基于图像处理技术的纱线毛羽检测应用研究[D].石家庄:河北科技大学,2011:2.ZHANG Jilei.Research on yarn hairiness detection based on image processing technology[D].Shijiazhuang:Hebei University of Science and Technology,2011:2.