基于PX-LBP和像素分类的装配体零件识别研究
|
摘要
针对机械产品装配维修诱导中零件和装配体的识别、监测问题,对装配体零件识别及装配监测进行了研究,对LBP算子进行了改进,提出了一种基于像素局部二值模式(PX-LBP)和像素分类的装配体零件识别及装配监测方法。首先将LBP算子与像素分类融合,提出了PX-LBP算子;然后对深度图像进行了PX-LBP特征提取,生成了训练集和测试集;最后训练随机森林分类器,并利用训练好的随机森林分类器实现了对测试集深度图像的像素分类,生成了像素预测图像,通过像素预测图像与标记图像对比实现了装配体零件的识别及装配过程的监测。研究结果表明:该方法对于模型深度图像的像素识别率可达到98.81%,对于真实装配体深度图像的像素识别率也可达到77.51%;该方法兼具了一定的实时性与鲁棒性,可用在装配维修诱导、装配监测和自动化装配邻域中。
Aiming at issue of parts recognition and assembly monitoring in assembly maintenance and guidance of mechanical products, a part recognition and assembly monitoring method based on Pixel Local Binary Pattern(PX-LBP) was proposed. Firstly, the classical LBP operator was merged with the pixel classification to propose an improved LBP operator, which is named PX-LBP operator. Secondly, PX-LBP features of the depth images were extracted. And training set and test set were obtained. Finally, the randomized decision forests classifier was trained. Then the pixel classification of the depth image of the test set was executed to get the pixel prediction image of depth images. The recognition of assembly parts was realized by comparing the RGB values of pixel prediction image and the corresponding color label image. The assembly process monitoring was realized by analyzing the number and position of pixels in pixel prediction image of each assembly part to determine the assembly error. Experiment results show that the method proposed in this paper has high accuracy, real-time and robust ability, and can be used in fields of assembly maintenance guiding, assembly monitoring and automatic assembly.
Aiming at issue of parts recognition and assembly monitoring in assembly maintenance and guidance of mechanical products, a part recognition and assembly monitoring method based on Pixel Local Binary Pattern(PX-LBP) was proposed. Firstly, the classical LBP operator was merged with the pixel classification to propose an improved LBP operator, which is named PX-LBP operator. Secondly, PX-LBP features of the depth images were extracted. And training set and test set were obtained. Finally, the randomized decision forests classifier was trained. Then the pixel classification of the depth image of the test set was executed to get the pixel prediction image of depth images. The recognition of assembly parts was realized by comparing the RGB values of pixel prediction image and the corresponding color label image. The assembly process monitoring was realized by analyzing the number and position of pixels in pixel prediction image of each assembly part to determine the assembly error. Experiment results show that the method proposed in this paper has high accuracy, real-time and robust ability, and can be used in fields of assembly maintenance guiding, assembly monitoring and automatic assembly.
引文
[1] CAO W, JIANG P Y, LU P, et al. Real-time data-driven monitoring in job-shop floor based on radio frequency identification [J]. The International Journal of Advanced Manufacturing Technology, 2017,92(5-8):2099-2120.
[2] YIN C H, WANG W, ZHAO G, et al. Research on rapid layout method of RFID devices for aircraft assembly monitoring [C]. Proceedings of 2016 6th International Conference on Information Technology for Manufacturing Systems, Prague:IEEE,2016.
[3] 周铨.一种过盈压装装配工序的过程监测系统[J].机电产品开发与创新,2018,31(3):84-86.
[4] 陈培,申红明,张会猛,等.基于ARM的线缆生产监控系统的设计[J].现代电子技术,2017,40(8):65-72.
[5] 汪嘉杰,王磊,范秀敏,等.基于视觉的航天电连接器的智能识别与装配引导[J],计算机集成制造系统,2017,23(11):2423-2430.
[6] OJALA T, PIETIKAINEN M, HARWOOD D. A comparative study of texture measures with classification based on featured distribu-tions [J]. Pattern Recognition, 1996,29(1):51-59.
[7] OJALA T, PIETIKAINEN M, MEMBER S, et al. Multiresolution gray-scale and rotation invariant texture classification with local binary patterns [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2002,24(7):971-987.
[8] LIAO S,ZHU X,LEI Z,et al. Learning multi-scale block local binary patterns for face recognition[J]. Lecture Notes in Computer Science,2007(4642):828-837.
[9] HEIKKILA M, PIETIKAINEN M, SCHMID C. Description of interestrions with local binary patterns[J]. Pattern Recognition, 2009,42(3):425-436.
[10] MATURANA D,MERY D,SOTO A. Learning discriminative local binary patterns for face recognition[C]. IEEE international conference on automatic face and gesture recognition,Santa Barbona:IEEE, 2011.
[11] HUSSAIN S U, NAPOLEON T,JURIE F . Face recognition using local quantized patterns[C]. British machine vision conference, Guildford: IEEE, 2012.
[12] 冀中,聂林红.基于抗噪声局部二值模式的纹理图像分类[J].计算机研究与发展,2016,53(5):1128-1135.
[13] 胡敏,腾文娣,王晓华,等.融合局部纹理和形状的人脸表情识别[J].电子与信息学报,2018,40(6):1338-1344.
[14] SHOTTON J, FITZGIBBON A, COOK M, et al. Real-time human pose recognition in parts from single depth images[J]. Computer Vision & Pattern Recognition, 2011,56(1):1297-1304.
[15] 田野,项世军.基于LBP和多层DCT的人脸活体检测算法[J].计算机研究展,2018,55(3):1128-1135.
[16] 李光梅,苟岩岩,刘果玲,等.易拉罐空罐缺陷智能轻测系统的设计[J].包装与食品机械,2017(6):33-36.
[17] 刘继忠,吴文虎,程乘,等.基于像素滤波和中值滤波的深度图像修复方法[J].光电子.激光,2018,29(5):539-544.
[18] XU J, DING X Q, WANG S J, et al. Background subtraction based on combination of texture, color and intensity[J]. ICSP: 2008 9th International Conference on Signal Processing, 2008(1):1400-1405.
[19] BREIMAN L. Randomized decision forests[J]. Machine Learning, 2001,45(1):5-32.
[20] BREIMAN L, FRIEDMAN J H, OLSHEN R A, et al. Classification and regression trees(CART)[J]. Biomtrics, 1984,40(3):358.
[2] YIN C H, WANG W, ZHAO G, et al. Research on rapid layout method of RFID devices for aircraft assembly monitoring [C]. Proceedings of 2016 6th International Conference on Information Technology for Manufacturing Systems, Prague:IEEE,2016.
[3] 周铨.一种过盈压装装配工序的过程监测系统[J].机电产品开发与创新,2018,31(3):84-86.
[4] 陈培,申红明,张会猛,等.基于ARM的线缆生产监控系统的设计[J].现代电子技术,2017,40(8):65-72.
[5] 汪嘉杰,王磊,范秀敏,等.基于视觉的航天电连接器的智能识别与装配引导[J],计算机集成制造系统,2017,23(11):2423-2430.
[6] OJALA T, PIETIKAINEN M, HARWOOD D. A comparative study of texture measures with classification based on featured distribu-tions [J]. Pattern Recognition, 1996,29(1):51-59.
[7] OJALA T, PIETIKAINEN M, MEMBER S, et al. Multiresolution gray-scale and rotation invariant texture classification with local binary patterns [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2002,24(7):971-987.
[8] LIAO S,ZHU X,LEI Z,et al. Learning multi-scale block local binary patterns for face recognition[J]. Lecture Notes in Computer Science,2007(4642):828-837.
[9] HEIKKILA M, PIETIKAINEN M, SCHMID C. Description of interestrions with local binary patterns[J]. Pattern Recognition, 2009,42(3):425-436.
[10] MATURANA D,MERY D,SOTO A. Learning discriminative local binary patterns for face recognition[C]. IEEE international conference on automatic face and gesture recognition,Santa Barbona:IEEE, 2011.
[11] HUSSAIN S U, NAPOLEON T,JURIE F . Face recognition using local quantized patterns[C]. British machine vision conference, Guildford: IEEE, 2012.
[12] 冀中,聂林红.基于抗噪声局部二值模式的纹理图像分类[J].计算机研究与发展,2016,53(5):1128-1135.
[13] 胡敏,腾文娣,王晓华,等.融合局部纹理和形状的人脸表情识别[J].电子与信息学报,2018,40(6):1338-1344.
[14] SHOTTON J, FITZGIBBON A, COOK M, et al. Real-time human pose recognition in parts from single depth images[J]. Computer Vision & Pattern Recognition, 2011,56(1):1297-1304.
[15] 田野,项世军.基于LBP和多层DCT的人脸活体检测算法[J].计算机研究展,2018,55(3):1128-1135.
[16] 李光梅,苟岩岩,刘果玲,等.易拉罐空罐缺陷智能轻测系统的设计[J].包装与食品机械,2017(6):33-36.
[17] 刘继忠,吴文虎,程乘,等.基于像素滤波和中值滤波的深度图像修复方法[J].光电子.激光,2018,29(5):539-544.
[18] XU J, DING X Q, WANG S J, et al. Background subtraction based on combination of texture, color and intensity[J]. ICSP: 2008 9th International Conference on Signal Processing, 2008(1):1400-1405.
[19] BREIMAN L. Randomized decision forests[J]. Machine Learning, 2001,45(1):5-32.
[20] BREIMAN L, FRIEDMAN J H, OLSHEN R A, et al. Classification and regression trees(CART)[J]. Biomtrics, 1984,40(3):358.