手工装配中微小金属件漏装模糊自适应视觉检测方法
|
摘要
针对手工装配作业中微小金属件装配区图像特征易受光照环境影响产生不规则动态变化的问题,定性分析了总体视场及微小金属件装配区灰度分布的影响因素,建立了不同装配特征灰度分布混杂区动态识别数学模型;通过BP神经网络学习总体视场及装配区灰度分布特征与不同装配特征之间的非线性关系,并将网络预测精度转化为灰度分布特征对装配特征的模糊隶属度函数,提出了一套基于灰度混杂区阈值模糊自适应调整的微小金属件漏装视觉检测方法。在某企业所生产的钢琴顶盖螺钉漏装机器视觉检测系统中的成功应用,结果表明所提方法对手工装配作业复杂光环境具有较好的鲁棒性和适应性。
Aiming at the irregular dynamic change of image feature on micro metal parts assembly area caused by complex illumination environment in the manual assembly operation,the factors affecting the gray distribution of the whole field of view and micro metal parts assembly area are qualitatively analyzed,and a mathematical model for the dynamic identification of the gray distribution hybrid area among different assembly features is established. Then the nonlinear relationship between global and local gray distribution characteristics and different assembly features is trained by BP neural network,and the network forecasting accuracy is transformed into the fuzzy membership function of gray distribution to assembly features. Finally based on fuzzy adaptive adjustment of gray level hybrid threshold,a machine vision inspection method for micro metal parts leakage in manual assembly operation is proposed. The successful application of the method proposed in a machine vision inspection system for piano cap screw leakage shows that it has good robustness and adaptability to complex light environment in manual assembly.
Aiming at the irregular dynamic change of image feature on micro metal parts assembly area caused by complex illumination environment in the manual assembly operation,the factors affecting the gray distribution of the whole field of view and micro metal parts assembly area are qualitatively analyzed,and a mathematical model for the dynamic identification of the gray distribution hybrid area among different assembly features is established. Then the nonlinear relationship between global and local gray distribution characteristics and different assembly features is trained by BP neural network,and the network forecasting accuracy is transformed into the fuzzy membership function of gray distribution to assembly features. Finally based on fuzzy adaptive adjustment of gray level hybrid threshold,a machine vision inspection method for micro metal parts leakage in manual assembly operation is proposed. The successful application of the method proposed in a machine vision inspection system for piano cap screw leakage shows that it has good robustness and adaptability to complex light environment in manual assembly.
引文
[1]WEN Jieding.Practice and analysis of life measuring device for motor valve[J].Applied Mechanics and Materials,2013(423/424/425/426):1972-1977.
[2]刘明周,蒋倩男,葛茂根.基于机器视觉的装配动作自动分割与识别[J].中国机械工程,2017,28(11):1346-1354.
[3]王耀南,陈铁健,贺振东,等.智能制造装备视觉检测控制方法综述[J].控制理论与应用,2015,32(3):273-286.
[4]杨庆华,陈亮,荀一,等.基于机器视觉的PCB裸板缺陷自动检测方法[J].中国机械工程,2012,23(22):2661-2666.
[5]PERNG Der-Baau,LIU Hsiao-Wei,CHANG Ching-Ching.Automated SMD LED inspection using machine vision[J].International Journal of Advanced Manufacturing Technology,2011,57(12):1065-1077.
[6]JEONG Young-Man,KIM Yong-Seok,KIM Seung-Soo,et al.Construction of an automation system for the inspection and packing processes of a screw/bolt production line[J].Journal of Mechanical Science and Technology,2013,27(6):1825-1834.
[7]HASHIM H S,PRABUWONO A S,ABDULLAH S N H S.Metal Parts Visual Inspection Based on Production Rules[J].Applied Mechanics and Materials,2011(110/111/112/113/114/115/116):4091-4095.
[8]黄凯奇,任伟强,谭铁牛.图像物体分类与检测算法综述[J].计算机学报,2014,37(6):1225-1240.
[9]周封,杨超,王晨光,等.基于随机Hough变换的复杂条件下圆检测与数目辨识[J].仪器仪表学报,2013,34(3):622-628.
[10]SUN Jun,SUN Qiao,BRIAN W Surgenor.An adaptable automated visual inspection scheme through online learning[J].International Journal of Advanced Manufacturing Technology,2012,59(5/6/7/8):655-667.
[11]LUO Shaofeng,ZOU Xiangjun,ZHANG Yuanqin,et al.Atarget detection method to mitigate interference of non-uniform illumination.2016 23rd International Conference on Mechatronics and Machine Vision in Practice(M2VIP)[C].Nanjing,China:IEEE:2016.
[12]魏巍.噪声和不均匀光照图像阈值分割技术研究[D].长春:吉林大学,2011.
[13]张槐祥,刘怀广,杨茂麟,等.复杂光照环境下的线材缺陷自适应分割方法[J].光电工程,2014(2):40-46.
[14]吴一全,孟天亮,吴诗婳.图像阈值分割方法研究进展20年(1994-2014)[J].数据采集与处理,2015,30(1):1-23.
[15]龙建武,申铉京,臧慧,等.高斯尺度空间下估计背景的自适应阈值分割算法[J].自动化学报,2014,40(8):1773-1782.
[2]刘明周,蒋倩男,葛茂根.基于机器视觉的装配动作自动分割与识别[J].中国机械工程,2017,28(11):1346-1354.
[3]王耀南,陈铁健,贺振东,等.智能制造装备视觉检测控制方法综述[J].控制理论与应用,2015,32(3):273-286.
[4]杨庆华,陈亮,荀一,等.基于机器视觉的PCB裸板缺陷自动检测方法[J].中国机械工程,2012,23(22):2661-2666.
[5]PERNG Der-Baau,LIU Hsiao-Wei,CHANG Ching-Ching.Automated SMD LED inspection using machine vision[J].International Journal of Advanced Manufacturing Technology,2011,57(12):1065-1077.
[6]JEONG Young-Man,KIM Yong-Seok,KIM Seung-Soo,et al.Construction of an automation system for the inspection and packing processes of a screw/bolt production line[J].Journal of Mechanical Science and Technology,2013,27(6):1825-1834.
[7]HASHIM H S,PRABUWONO A S,ABDULLAH S N H S.Metal Parts Visual Inspection Based on Production Rules[J].Applied Mechanics and Materials,2011(110/111/112/113/114/115/116):4091-4095.
[8]黄凯奇,任伟强,谭铁牛.图像物体分类与检测算法综述[J].计算机学报,2014,37(6):1225-1240.
[9]周封,杨超,王晨光,等.基于随机Hough变换的复杂条件下圆检测与数目辨识[J].仪器仪表学报,2013,34(3):622-628.
[10]SUN Jun,SUN Qiao,BRIAN W Surgenor.An adaptable automated visual inspection scheme through online learning[J].International Journal of Advanced Manufacturing Technology,2012,59(5/6/7/8):655-667.
[11]LUO Shaofeng,ZOU Xiangjun,ZHANG Yuanqin,et al.Atarget detection method to mitigate interference of non-uniform illumination.2016 23rd International Conference on Mechatronics and Machine Vision in Practice(M2VIP)[C].Nanjing,China:IEEE:2016.
[12]魏巍.噪声和不均匀光照图像阈值分割技术研究[D].长春:吉林大学,2011.
[13]张槐祥,刘怀广,杨茂麟,等.复杂光照环境下的线材缺陷自适应分割方法[J].光电工程,2014(2):40-46.
[14]吴一全,孟天亮,吴诗婳.图像阈值分割方法研究进展20年(1994-2014)[J].数据采集与处理,2015,30(1):1-23.
[15]龙建武,申铉京,臧慧,等.高斯尺度空间下估计背景的自适应阈值分割算法[J].自动化学报,2014,40(8):1773-1782.