用于卷积神经网络图像预处理的目标中心化算法
|
摘要
为解决工业生产中对不同零件进行自动分类的问题,提出一种基于卷积神经网络的模式识别算法,对29种不同尺寸的螺丝、螺母和垫片进行分类。首先采集待分类零件的图像数据,通过数据增强得到数据集,然后设计一种简化的卷积神经网络。提出一种对图像中的目标位置进行中心化的图像预处理算法,它能够提取图像中目标所在的区域并将其移动到图像中心位置。研究结果表明,与不采用目标中心化算法的传统方法相比,总体准确率从97.59%提升至99.96%,具有最低准确率的零件的准确率从85.83%提升至99.67%。使用卷积神经网络对背景纯净且目标明显的图像进行分类时,使用本文提出的目标中心化算法进行图像预处理能够显著提高网络的识别准确率。
To solve the problem of classifying different parts automatically in industrial production, a pattern recognition algorithm based on convolutional neural network was raised and 29 different sizes of screws, nuts and washers were classified. Firstly, image data of the parts that were going to be classified were collected, and the dataset was created by data augmentation. Then, a simplified convolutional neural network was designed. An image preprocessing algorithm to centralize the position of the target in the image was raised, which can extract the target area in the image and move it to the center of the image. The results show that compared with traditional method without target-centralization algorithm,the total error is raised from 97.69% to 99.96, and the accuracy of the part which has the lowest accuracy is raised from85.83% to 99.67%. When convolutional neural network is used to classify images which has pure background and obvious object, using the target-centralization algorithm raised in this paper to preprocess the images can improve the accuracy of the network significantly.
To solve the problem of classifying different parts automatically in industrial production, a pattern recognition algorithm based on convolutional neural network was raised and 29 different sizes of screws, nuts and washers were classified. Firstly, image data of the parts that were going to be classified were collected, and the dataset was created by data augmentation. Then, a simplified convolutional neural network was designed. An image preprocessing algorithm to centralize the position of the target in the image was raised, which can extract the target area in the image and move it to the center of the image. The results show that compared with traditional method without target-centralization algorithm,the total error is raised from 97.69% to 99.96, and the accuracy of the part which has the lowest accuracy is raised from85.83% to 99.67%. When convolutional neural network is used to classify images which has pure background and obvious object, using the target-centralization algorithm raised in this paper to preprocess the images can improve the accuracy of the network significantly.
引文
[1]何晓阳,徐惠钢,谢启.基于LabVIEW与BP神经网络的零件识别系统[J].仪表技术与传感器,2017(1):119-122.HE Xiaoyang,XU Huigang,XIE Qi.Recognition system of parts based on LabVIEW and BP neural network[J].Instrument Technique and Sensor,2017(1):119-122.
[2]何泽强.基于机器视觉的工业机器人分拣系统设计[D].哈尔滨:哈尔滨工业大学信息与电气工程学院,2016:1.HE Zeqiang.Design of industrial robot sorting system based on machine vision[D].Harbin:Harbin Institute of Technology.School of Information and Electrical Engineering,2016:1.
[3]GOODFELLOW I,BENGIO Y,COURVILLE A.Deep learning[M].Cambridge,MA:The MIT Press,2016:3.
[4]司小婷,吴文江,孙一兰.基于视觉的零件识别和定位[J].组合机床与自动化加工技术,2016(10):70-73.SI Xiaoting,WU Wenjiang,SUN Yilan.The identification and positioning of parts based on machine vision[J].Modular Machine Tool&Automatic Manufacturing Technique,2016(10):70-73.
[5]刘振宇,李中生,赵雪,等.基于机器视觉的工业机器人分拣技术研究[J].制造业自动化,2013(17):25-30.LIU Zhenyu,LI Zhongsheng,ZHAO Xue,et al.Research of sorting technology based on industrial robot of machine vision[J].Manufacturing Automation,2013(17):25-30.
[6]吴益红,许钢,江娟娟,等.基于LBP和SVM的工件图像特征识别研究[J].重庆理工大学学报,2016,30(1):77-84.WU Yihong,XU Gang,JIANG Juanjuan,et al.Research on workpiece image feature recognition based on LBP and SVM[J].Journal of Chongqing University of Technology(Natural Science),2016,30(1):77-84.
[7]冯长建,吴斌,罗跃纲.混合KPCA和SVM的机械零件形状识别方法研究[J].机械制造与自动化,2016(4):132-134.FENG Changjian,WU Bin,LUO Yuegang.Research on shape recognition of mechanical parts based on hybrid KPCA and SVM[J].Machine Building&Automation,2016(4):132-134.
[8]李春,李琳,邹焱飚,等.基于视觉的焊接工件在线识别与分类算法研究[J].价值工程,2016,35(4):97-101.LI Chun,LI Lin,ZOU Yanbiao,et al.Research on on-line recognition and classification of weldment based on machine vision[J].Value Engineering,2016,35(4):97-101.
[9]HINTON G E,SALAKHUTDINOV R R.Reducing the dimensionality of data with neural networks[J].Science,2006,313(5786):504.
[10]赵鹏,王斐,刘慧婷,等.基于深度学习的手绘草图识别[J].四川大学学报(工程科学版),2016,48(3):94-99.ZHAO Peng,WANG Fei,LIU Huiting,et al.Sketch recognition using deep learning[J].Journal of Sichuan University(Engineering Science Edition),2016,48(3):94-99.
[11]LéCUN Y,BOTTOU L,BENGIO Y,et al.Gradient-based learning applied to document recognition[J].Proceedings of the IEEE,2001,86(11):2278-2324.
[12]杨国国,鲍一丹,刘子毅.基于图像显著性分析与卷积神经网络的茶园害虫定位与识别[J].农业工程学报,2017,33(6):156-162.YANG Guoguo,BAO Yidan,LIU Ziyi.Localization and recognition of pests in tea plantation based on image saliency analysis and convolutional neural network[J].Transactions of the Chinese Society of Agricultural Engineering,2017,33(6):156-162.
[13]周云成,许童羽,郑伟,等.基于深度卷积神经网络的番茄主要器官分类识别方法[J].农业工程学报,2017,33(15):219-226.ZHOU Yuncheng,XU Tongyu,ZHENG Wei,et al.Classification and recognition approaches of tomato main organs based on DCNN[J].Transactions of the Chinese Society of Agricultural Engineering,2017,33(15):219-226.
[14]SERMANET P,EIGEN D,ZHANG X,et al.Over Feat:integrated recognition,localization and detection using convolutional networks[EB/OL].[2013-12-21].https://arxiv.org/abs/1312.6229.
[15]HE K,ZHANG X,REN S,et al.Deep Residual Learning for Image Recognition[C]//Computer Vision and Pattern Recognition.Piscataway,NJ:IEEE,2016:770-778.
[16]高学,王有旺.基于CNN和随机弹性形变的相似手写汉字识别[J].华南理工大学学报(自然科学版),2014,42(1):72-76.GAO Xue,WANG Youwang.Recognition of similar handwritten Chinese characters based on CNN and random elastic deformation[J].Journal of South China University of Technology(Natural Science),2014,42(1):72-76.
[17]GLOROT X,BORDES A,BENGIO Y,et al.Deep sparse rectifier neural networks[C]//International Conference on Artificial Intelligence and Statistics.Cambridge,MA:The MITPress,2012:315-323.
[18]ZHOU Y T,CHELLAPPA R.Computation of optical flow using a neural network[C]//IEEE International Conference on Neural Networks.Piscataway,NJ:IEEE,1988:71-78.
[19]KRIZHEVSKY A,SUTSKEVER I,HINTON G E.ImageNet classification with deep convolutional neural networks[J].Communications of the ACM,2017,60(2):2012.
[20]SIMONYAN K,ZISSERMAN A.Very Deep Convolutional networks for large-scale image recognition[EB/OL].[2014-10-15].https://arxiv.org/abs/1409.1556.
[21]SRIVASTAVA N,HINTON G,KRIZHEVSKY A,et al.Dropout:a simple way to prevent neural networks from overfitting[J].Journal of Machine Learning Research,2014,15(1):1929-1958.
[22]KINGMA D P,BA J.Adam:A method for stochastic optimization[EB/OL].[2014-12-22].https://arxiv.org/abs/1412.6980.
[23]IOFFE S,SZEGEDY C.Batch normalization:accelerating deep network training by reducing internal covariate shift[EB/OL].[2015-01-11].https://arxiv.org/abs/1502.0316.
[2]何泽强.基于机器视觉的工业机器人分拣系统设计[D].哈尔滨:哈尔滨工业大学信息与电气工程学院,2016:1.HE Zeqiang.Design of industrial robot sorting system based on machine vision[D].Harbin:Harbin Institute of Technology.School of Information and Electrical Engineering,2016:1.
[3]GOODFELLOW I,BENGIO Y,COURVILLE A.Deep learning[M].Cambridge,MA:The MIT Press,2016:3.
[4]司小婷,吴文江,孙一兰.基于视觉的零件识别和定位[J].组合机床与自动化加工技术,2016(10):70-73.SI Xiaoting,WU Wenjiang,SUN Yilan.The identification and positioning of parts based on machine vision[J].Modular Machine Tool&Automatic Manufacturing Technique,2016(10):70-73.
[5]刘振宇,李中生,赵雪,等.基于机器视觉的工业机器人分拣技术研究[J].制造业自动化,2013(17):25-30.LIU Zhenyu,LI Zhongsheng,ZHAO Xue,et al.Research of sorting technology based on industrial robot of machine vision[J].Manufacturing Automation,2013(17):25-30.
[6]吴益红,许钢,江娟娟,等.基于LBP和SVM的工件图像特征识别研究[J].重庆理工大学学报,2016,30(1):77-84.WU Yihong,XU Gang,JIANG Juanjuan,et al.Research on workpiece image feature recognition based on LBP and SVM[J].Journal of Chongqing University of Technology(Natural Science),2016,30(1):77-84.
[7]冯长建,吴斌,罗跃纲.混合KPCA和SVM的机械零件形状识别方法研究[J].机械制造与自动化,2016(4):132-134.FENG Changjian,WU Bin,LUO Yuegang.Research on shape recognition of mechanical parts based on hybrid KPCA and SVM[J].Machine Building&Automation,2016(4):132-134.
[8]李春,李琳,邹焱飚,等.基于视觉的焊接工件在线识别与分类算法研究[J].价值工程,2016,35(4):97-101.LI Chun,LI Lin,ZOU Yanbiao,et al.Research on on-line recognition and classification of weldment based on machine vision[J].Value Engineering,2016,35(4):97-101.
[9]HINTON G E,SALAKHUTDINOV R R.Reducing the dimensionality of data with neural networks[J].Science,2006,313(5786):504.
[10]赵鹏,王斐,刘慧婷,等.基于深度学习的手绘草图识别[J].四川大学学报(工程科学版),2016,48(3):94-99.ZHAO Peng,WANG Fei,LIU Huiting,et al.Sketch recognition using deep learning[J].Journal of Sichuan University(Engineering Science Edition),2016,48(3):94-99.
[11]LéCUN Y,BOTTOU L,BENGIO Y,et al.Gradient-based learning applied to document recognition[J].Proceedings of the IEEE,2001,86(11):2278-2324.
[12]杨国国,鲍一丹,刘子毅.基于图像显著性分析与卷积神经网络的茶园害虫定位与识别[J].农业工程学报,2017,33(6):156-162.YANG Guoguo,BAO Yidan,LIU Ziyi.Localization and recognition of pests in tea plantation based on image saliency analysis and convolutional neural network[J].Transactions of the Chinese Society of Agricultural Engineering,2017,33(6):156-162.
[13]周云成,许童羽,郑伟,等.基于深度卷积神经网络的番茄主要器官分类识别方法[J].农业工程学报,2017,33(15):219-226.ZHOU Yuncheng,XU Tongyu,ZHENG Wei,et al.Classification and recognition approaches of tomato main organs based on DCNN[J].Transactions of the Chinese Society of Agricultural Engineering,2017,33(15):219-226.
[14]SERMANET P,EIGEN D,ZHANG X,et al.Over Feat:integrated recognition,localization and detection using convolutional networks[EB/OL].[2013-12-21].https://arxiv.org/abs/1312.6229.
[15]HE K,ZHANG X,REN S,et al.Deep Residual Learning for Image Recognition[C]//Computer Vision and Pattern Recognition.Piscataway,NJ:IEEE,2016:770-778.
[16]高学,王有旺.基于CNN和随机弹性形变的相似手写汉字识别[J].华南理工大学学报(自然科学版),2014,42(1):72-76.GAO Xue,WANG Youwang.Recognition of similar handwritten Chinese characters based on CNN and random elastic deformation[J].Journal of South China University of Technology(Natural Science),2014,42(1):72-76.
[17]GLOROT X,BORDES A,BENGIO Y,et al.Deep sparse rectifier neural networks[C]//International Conference on Artificial Intelligence and Statistics.Cambridge,MA:The MITPress,2012:315-323.
[18]ZHOU Y T,CHELLAPPA R.Computation of optical flow using a neural network[C]//IEEE International Conference on Neural Networks.Piscataway,NJ:IEEE,1988:71-78.
[19]KRIZHEVSKY A,SUTSKEVER I,HINTON G E.ImageNet classification with deep convolutional neural networks[J].Communications of the ACM,2017,60(2):2012.
[20]SIMONYAN K,ZISSERMAN A.Very Deep Convolutional networks for large-scale image recognition[EB/OL].[2014-10-15].https://arxiv.org/abs/1409.1556.
[21]SRIVASTAVA N,HINTON G,KRIZHEVSKY A,et al.Dropout:a simple way to prevent neural networks from overfitting[J].Journal of Machine Learning Research,2014,15(1):1929-1958.
[22]KINGMA D P,BA J.Adam:A method for stochastic optimization[EB/OL].[2014-12-22].https://arxiv.org/abs/1412.6980.
[23]IOFFE S,SZEGEDY C.Batch normalization:accelerating deep network training by reducing internal covariate shift[EB/OL].[2015-01-11].https://arxiv.org/abs/1502.0316.