基于图像光流的轴承滚子表面缺陷检测
|
摘要
针对轴承滚子表面缺陷人工检测效率低、误检率高的问题,以及现有机器视觉检测方法存在的不足。基于上述情况,提出了一种基于图像光流的轴承滚子表面缺陷的检测方法。首先设计了一套图像采集装置,用于获取轴承滚子表面图像;其次采用全局和局部相结合的平滑策略,建立光流误差估计模型,同时为了增强缺陷位置的光流边界,在此模型中引入各项异性的扩散张量;在光流求解策略上,先对原始图像进行奇异值分解,消除灰度值异常对光流计算的影响,再引入金字塔分层细化方法,提高光流计算的准确性;最后根据光流生成的伪彩色图像,完成对缺陷区域的粗略定位,分割出缺陷区域。实验表明,所提方法在轴承滚子表面缺陷检测方面的可行性和有效性。
In the domain of bearing roller surface defect detection,both the human visual inspection and machine vision have problems,including low efficiency,high error rate,etc. To solve these problems,a detection method of bearing roller surface defects based on image optical flow is proposed in this article. First,a set of image acquisition devices is designed to obtain the image of bearing roller surface. Secondly,the combination of global and local smoothing strategy is adopted to formulate the estimation error model. At the same time,to enhance the optical flow boundary of the defect position,the diffusion tensor of the anisotropy is introduced in the model. For the solution strategy of optical flow,this paper carries out the singular value decomposition on the original image to eliminate the influence of gray value anomalies. The pyramid stratification method is used to improve the accuracy of optical flow calculations. Finally,according to the pseudo-color image generated by the optical flow,a rough location of the defect area is completed to segment the defect area. Experimental results show that the proposed method is feasible and effective on detecting the bearing roller surface defect.
In the domain of bearing roller surface defect detection,both the human visual inspection and machine vision have problems,including low efficiency,high error rate,etc. To solve these problems,a detection method of bearing roller surface defects based on image optical flow is proposed in this article. First,a set of image acquisition devices is designed to obtain the image of bearing roller surface. Secondly,the combination of global and local smoothing strategy is adopted to formulate the estimation error model. At the same time,to enhance the optical flow boundary of the defect position,the diffusion tensor of the anisotropy is introduced in the model. For the solution strategy of optical flow,this paper carries out the singular value decomposition on the original image to eliminate the influence of gray value anomalies. The pyramid stratification method is used to improve the accuracy of optical flow calculations. Finally,according to the pseudo-color image generated by the optical flow,a rough location of the defect area is completed to segment the defect area. Experimental results show that the proposed method is feasible and effective on detecting the bearing roller surface defect.
引文
[1]夏新涛,常振,叶亮,等.滚动轴承性能稳定性及可靠性分析[J].仪器仪表学报,2017,38(6):1421-1431.XIA X T,CHENG ZH,YE L,et al.Stability and reliability analysis of rolling bearing performance[J].Chinese Journal of Scientific Instrument,2017,38(6):1421-1431.
[2]梁子千,玄文博,王婷,等.基于二维阻抗特征的管道环焊缝缺陷涡流检测[J].仪器仪表学报,2017,38(9):2138-2145.LIANG Z Q,XUAN W B,WANG T,et al.Eddy current NDt for the cracks of girth welds of pipes based on 2D impedance characteristics[J].Chinese Journal of Scientific Instrument,2017,38(9):2138-2145.
[3]陈振华,郑志远,卢超.不锈钢焊缝中超声传播特性及TOFD检测方法研究[J].电子测量与仪器学报,2017,31(7):1129-1136.CHEN ZH H,ZHENG ZH Y,LU CH.Research on wave propagation characteristics in austenite stainless steel weld and ultrasonic TOFD testing technique[J].Journal of Electronic Measurement and Instrumentation,2017,31(7):1129-1136.
[4]王姮,卜燕,张华,等.基于分形维数的磁痕图像缺陷检测[J].计算机应用研究,2015,32(2):603-605.WANG H,BU L,ZHANG H,et al.Defect detection of magnetic particle indication image based on fractal dimension[J].Application Research of Computers,2015,32(2):603-605.
[5]苏俊宏,刘胜利.圆柱型高精密零件表面缺陷检测及形貌分析[J].激光与光电子学进展,2014,51(4):150-154.SU J H,LIU SH L.Measure system of surface flaw and morphology analysis of cylindrical high precision parts[J].Laser&Optoelectronics Progress,2014,51(4):150-154.
[6]孙虹,王仲,付鲁华,等.一种新型钢球表面检测方法的分析与系统实现[J].机械科学与技术,2016,35(1):118-121.SUN H,WANG ZH,FU L H,et al.Analysis and implementation of novel method for steel ball surface detection[J].Mechanical Science and Technology for Aerospace Engineering,2016,35(1):118-121.
[7]刘青,张晓晖,黄军勤.基于机器视觉的钢球表面缺陷快速提取算法[J].激光杂志,2014,35(9):58-61.LIU Q,ZHANG X H,HUANG J Q.The rapid extracting algorithm for steel ball surface defect based on machine vision[J].Laser Journal,2014,35(9):58-61.
[8]陈丹阳,曹丽,林一高,等.轴承圆锥滚子外观缺陷检测研究[J].机电工程,2015,32(8):1084-1087.CHEN D Y,CAO L,LIN Y G,et al.Appearance defect detection of tapered roller[J].Journal of Mechanical&Electrical Engineering,2015,32(8):58-61.
[9]LIU R,LU Y.Infrared target tracking in multiple feature pseudo-color image with kernel density estimation[J].Infrared Physics&Technology,2012,55(6):505-512.
[10]AKSHAY S.Single moving object detection and tracking using Horn-Schunk optical flow method[J].International Journal of Applied Engineering Research,2015,10(11):30135-30152.
[11]NUSRAT S,REMUS B.Optimal filter estimation for Lucas-Kanade optical flow[J].Sensors,2012,12(9):12694-12709.
[12]JHA D K,GUPTA B,LAMBA S S.l2-norm-based prior for haze-removal from single image[J].IET Computer Vision,2016,10(5):331-341.
[13]MARKOPOULOS P P,KUNDU S,CHAMADIA S,et al.Efficient L1-norm principal-component analysis via bit flipping[J].IEEE Transactions on Signal Processing,2017,65(8):4252-4264.
[14]高银,云利军,石俊生,等.基于各向异性高斯滤波的暗原色理论雾天彩色图像增强算法[J].计算机辅助设计与图形学学报,2015,27(9):1701-1706.GAO Y,YUN L J,SHI J SH,et al.Enhancement dark channel algorithm of color fog image based on the anisotropic Gaussian filtering[J].Journal of ComputerAided Design&Computer Graphics,2015,27(9):1701-1706.
[15]廖建尚,王立国,郝思媛.基于双边滤波和空间邻域信息的高光谱图像分类方法[J].农业机械学报,2017,48(8):140-146.LIAO J SH,WANG L G,HAO S Y.Hyperspectral image classification method combined with bilateral filtering and pixel neighborhood information[J].Transactions of the Chinese Society for Agricultural Machinery,2017,48(8):140-146.
[16]任克强,梁亮亮,喻玲娟.SVD与HVS相结合的小波域数字图像鲁棒水印[J].电子测量与仪器学报,2017,31(6):869-875.REN K Q,LIANG L L,YU L J.Robust digital imagewatermarking in wavelet domain based on SVD and HVS[J].Journal of Electronic Measurement and Instrumentation,2017,31(6):869-875.
[17]申森,李艾华,姚良,等.基于小波包和Niblack法的枪号图像二值化算法[J].光子学报,2013,42(3):354-358.SHEN S,LI AI H,YAO L,et al.Gun code binary image algorithm based on wavelet packet and Niblack method[J].Acta Photonica Sinica,2013,42(3):354-358.
[2]梁子千,玄文博,王婷,等.基于二维阻抗特征的管道环焊缝缺陷涡流检测[J].仪器仪表学报,2017,38(9):2138-2145.LIANG Z Q,XUAN W B,WANG T,et al.Eddy current NDt for the cracks of girth welds of pipes based on 2D impedance characteristics[J].Chinese Journal of Scientific Instrument,2017,38(9):2138-2145.
[3]陈振华,郑志远,卢超.不锈钢焊缝中超声传播特性及TOFD检测方法研究[J].电子测量与仪器学报,2017,31(7):1129-1136.CHEN ZH H,ZHENG ZH Y,LU CH.Research on wave propagation characteristics in austenite stainless steel weld and ultrasonic TOFD testing technique[J].Journal of Electronic Measurement and Instrumentation,2017,31(7):1129-1136.
[4]王姮,卜燕,张华,等.基于分形维数的磁痕图像缺陷检测[J].计算机应用研究,2015,32(2):603-605.WANG H,BU L,ZHANG H,et al.Defect detection of magnetic particle indication image based on fractal dimension[J].Application Research of Computers,2015,32(2):603-605.
[5]苏俊宏,刘胜利.圆柱型高精密零件表面缺陷检测及形貌分析[J].激光与光电子学进展,2014,51(4):150-154.SU J H,LIU SH L.Measure system of surface flaw and morphology analysis of cylindrical high precision parts[J].Laser&Optoelectronics Progress,2014,51(4):150-154.
[6]孙虹,王仲,付鲁华,等.一种新型钢球表面检测方法的分析与系统实现[J].机械科学与技术,2016,35(1):118-121.SUN H,WANG ZH,FU L H,et al.Analysis and implementation of novel method for steel ball surface detection[J].Mechanical Science and Technology for Aerospace Engineering,2016,35(1):118-121.
[7]刘青,张晓晖,黄军勤.基于机器视觉的钢球表面缺陷快速提取算法[J].激光杂志,2014,35(9):58-61.LIU Q,ZHANG X H,HUANG J Q.The rapid extracting algorithm for steel ball surface defect based on machine vision[J].Laser Journal,2014,35(9):58-61.
[8]陈丹阳,曹丽,林一高,等.轴承圆锥滚子外观缺陷检测研究[J].机电工程,2015,32(8):1084-1087.CHEN D Y,CAO L,LIN Y G,et al.Appearance defect detection of tapered roller[J].Journal of Mechanical&Electrical Engineering,2015,32(8):58-61.
[9]LIU R,LU Y.Infrared target tracking in multiple feature pseudo-color image with kernel density estimation[J].Infrared Physics&Technology,2012,55(6):505-512.
[10]AKSHAY S.Single moving object detection and tracking using Horn-Schunk optical flow method[J].International Journal of Applied Engineering Research,2015,10(11):30135-30152.
[11]NUSRAT S,REMUS B.Optimal filter estimation for Lucas-Kanade optical flow[J].Sensors,2012,12(9):12694-12709.
[12]JHA D K,GUPTA B,LAMBA S S.l2-norm-based prior for haze-removal from single image[J].IET Computer Vision,2016,10(5):331-341.
[13]MARKOPOULOS P P,KUNDU S,CHAMADIA S,et al.Efficient L1-norm principal-component analysis via bit flipping[J].IEEE Transactions on Signal Processing,2017,65(8):4252-4264.
[14]高银,云利军,石俊生,等.基于各向异性高斯滤波的暗原色理论雾天彩色图像增强算法[J].计算机辅助设计与图形学学报,2015,27(9):1701-1706.GAO Y,YUN L J,SHI J SH,et al.Enhancement dark channel algorithm of color fog image based on the anisotropic Gaussian filtering[J].Journal of ComputerAided Design&Computer Graphics,2015,27(9):1701-1706.
[15]廖建尚,王立国,郝思媛.基于双边滤波和空间邻域信息的高光谱图像分类方法[J].农业机械学报,2017,48(8):140-146.LIAO J SH,WANG L G,HAO S Y.Hyperspectral image classification method combined with bilateral filtering and pixel neighborhood information[J].Transactions of the Chinese Society for Agricultural Machinery,2017,48(8):140-146.
[16]任克强,梁亮亮,喻玲娟.SVD与HVS相结合的小波域数字图像鲁棒水印[J].电子测量与仪器学报,2017,31(6):869-875.REN K Q,LIANG L L,YU L J.Robust digital imagewatermarking in wavelet domain based on SVD and HVS[J].Journal of Electronic Measurement and Instrumentation,2017,31(6):869-875.
[17]申森,李艾华,姚良,等.基于小波包和Niblack法的枪号图像二值化算法[J].光子学报,2013,42(3):354-358.SHEN S,LI AI H,YAO L,et al.Gun code binary image algorithm based on wavelet packet and Niblack method[J].Acta Photonica Sinica,2013,42(3):354-358.