基于混合蛙跳优化的采摘机器人相机标定方法
|
摘要
针对采摘机器人领域传统的张正友相机标定方法存在对相机模型参数初值敏感和标定结果不稳定等问题,提出一种基于改进混合蛙跳和LM算法的相机标定方法。该方法把相机标定划分为两步:(1)以混合蛙跳优化为工具,求出相机模型参数的初始值,避免传统张正友相机标定方法直接求取相机模型的参数初值所带来的初值敏感问题。(2)以改进LM算法对第1步求出的相机模型参数初值进行非线性优化求精,避免张正友相机标定方法须求取相机模型优化参数的雅可比矩阵,从而导致标定结果不稳定的问题。采用Open CV编写采摘机器人双目视觉标定系统,分别对传统张正友相机标定方法、基于遗传算法的相机标定方法、基于标准混合蛙跳算法的相机标定方法和本文相机标定方法进行相机标定试验。试验结果表明:本文相机标定方法所获得的左相机焦距的绝对误差为0. 065～0. 506 mm、相对误差为1. 899%～12. 652%,平面靶标图像特征点的平均像素误差为0. 166～0. 175像素;右相机焦距的绝对误差为0. 083～0. 360 mm、相对误差为2. 429%～11. 484%,平面靶标图像特征点的平均像素误差为0. 103～0. 114像素;双目相机之间距离的绝对误差为1. 866～2. 789 mm、相对误差为3. 209%～4. 874%。以上参数精度及收敛速度和稳定性均优于其他相机标定方法,从而验证了该方法所获得的相机标定参数具有较高的准确性和可靠性。
Due to the traditional Zhang Zhengyou's camera calibration method of picking robot existed the problems such as sensitive to initial value of camera model parameters and instability of calibration results,a camera calibration method based on improved shuffled frog leaping optimization and LM algorithm was proposed. The camera calibration was divided into two steps: the first step,calculating the initial values of the parameters of camera model with the shuffled frog leaping optimization,which avoided the sensitivity to the initial value of the camera model parameters that was directly calculated with the traditional Zhang Zhengyou's camera calibration method; the second step,refining the initial values of the parameters of camera model that calculated in the first step with improved nonlinear optimization LM algorithm,which avoided must obtaining the Jacobi matrix to optimize the parameters of the camera model with the Zhang Zhengyou's camera calibration method,which led to the instability of the calibration results. And the binocular vision calibration system of the picking robot was developed by OpenCV. Thecamera calibration experiments were carried out on the traditional Zhang Zhengyou's camera calibration method,the camera calibration method based on genetic algorithm,the camera calibration method based on shuffled frog leaping optimization algorithm and the camera calibration method. The test results showed that the absolute error of the left camera focal length was 0. 065 ～ 0. 100 mm,the relative error of the left camera focal length was 1. 899% ～ 12. 652%,the average pixel error of the left plane target image was0. 166 ～ 0. 175 pixel,the absolute error of the right camera focal length was 0. 083 ～ 0. 360 mm,the relative error of the right camera focal length was 2. 429% ～ 11. 484%,the average pixel error of the right plane target image was 0. 103 ～ 0. 114 pixel and the absolute error of distance of binocular camera was 1. 866 ～ 2. 789 mm,the relative error of the distance between the binocular camera was 3. 209% ～4. 874%,the convergence speed and stability,which were obtained by the camera calibration method,were all better than the other camera calibration methods in the above. So,these test results verified the calibration parameters obtained by the method had high accuracy and reliability.
Due to the traditional Zhang Zhengyou's camera calibration method of picking robot existed the problems such as sensitive to initial value of camera model parameters and instability of calibration results,a camera calibration method based on improved shuffled frog leaping optimization and LM algorithm was proposed. The camera calibration was divided into two steps: the first step,calculating the initial values of the parameters of camera model with the shuffled frog leaping optimization,which avoided the sensitivity to the initial value of the camera model parameters that was directly calculated with the traditional Zhang Zhengyou's camera calibration method; the second step,refining the initial values of the parameters of camera model that calculated in the first step with improved nonlinear optimization LM algorithm,which avoided must obtaining the Jacobi matrix to optimize the parameters of the camera model with the Zhang Zhengyou's camera calibration method,which led to the instability of the calibration results. And the binocular vision calibration system of the picking robot was developed by OpenCV. Thecamera calibration experiments were carried out on the traditional Zhang Zhengyou's camera calibration method,the camera calibration method based on genetic algorithm,the camera calibration method based on shuffled frog leaping optimization algorithm and the camera calibration method. The test results showed that the absolute error of the left camera focal length was 0. 065 ～ 0. 100 mm,the relative error of the left camera focal length was 1. 899% ～ 12. 652%,the average pixel error of the left plane target image was0. 166 ～ 0. 175 pixel,the absolute error of the right camera focal length was 0. 083 ～ 0. 360 mm,the relative error of the right camera focal length was 2. 429% ～ 11. 484%,the average pixel error of the right plane target image was 0. 103 ～ 0. 114 pixel and the absolute error of distance of binocular camera was 1. 866 ～ 2. 789 mm,the relative error of the distance between the binocular camera was 3. 209% ～4. 874%,the convergence speed and stability,which were obtained by the camera calibration method,were all better than the other camera calibration methods in the above. So,these test results verified the calibration parameters obtained by the method had high accuracy and reliability.
引文
[1] YAO H,ZHANG Z. Research of camera calibration based on genetic algorithm BP neural network[C]∥IEEE International Conference on Information and Automation,2017:350-355.
[2] FHR G,JUNG C R. Camera self-calibration based on nonlinear optimization and applications in surveillance systems[J].IEEE Transactions on Circuits&Systems for Video Technology,2017,27(5):1132-1142.
[3] PITCHANDI N,SUBRAMANIAN S P. GA-based camera calibration for vision-assisted robotic assembly system[J]. Iet Computer Vision,2017,11(1):50-59.
[4] ZHANG Y,ZHOU L,LIU H,et al. A flexible online camera calibration using line segments[J]. Journal of Sensors,2016:Article ID 2802343.
[5]游江,唐力伟,邓士杰.基于改进遗传模拟退火的相机标定方法[J].计算机工程与设计,2017,38(3):819-824.YOU Jiang,TANG Liwei,DENG Shijie. Camera calibration method based on genetic algorithm and simulated annealing[J].Computer Engineering and Design,2017,38(3):819-824.(in Chinese)
[6]柯丰恺,陈幼平,谢经明,等.基于凸松弛优化算法的相机内外参数标定[J].中国机械工程,2016,27(5):634-638.KE Fengkai,CHEN Youping,XIE Jingming,et al. Internal and external parameter calibrations of a camera based on convex relaxation optimization algorithm[J]. China Mechanical Engineering,2016,27(5):634-638.(in Chinese)
[7]刘艳,李腾飞.对张正友相机标定法的改进研究[J].光学技术,2014,40(6):565-570.LIU Yan,LI Tengfei. Reaserch of the improvement of Zhang's camera calibration method[J]. Optical Technique,2014,40(6):565-570.(in Chinese)
[8] D'EMILIA G,GASBARRO D D. Review of techniques for 2D camera calibration suitable for industrial vision systems[C]∥Journal of Physics:Conference Series,2017,841:012030.
[9] ZHANG Z. A flexible new technique for camera calibration[J]. Tpami,2000,22(11):1330-1334.
[10] GARY B,ADRIAN K. Learning Open CV[M]. California:O'Reilly Media,2008.
[11] EUSUFF M,LANSEY K,PASHA F. Shuffled frog-leaping algorithm:a memetic meta-heuristic for discrete optimization[J].Engineering Optimization,2006,38(2):129-154.
[12] SARKHEYLI A,ZAIN A M,SHARIF S. The role of basic,modified and hybrid shuffled frog leaping algorithm on optimization problems:a review[J]. Soft Computing,2015,19(7):1-28.
[13] JADIDOLESLAM M,EBRAHIMI A. Reliability constrained generation expansion planning by a modified shuffled frog leaping algorithm[J]. International Journal of Electrical Power&Energy Systems,2015,64:743-751.
[14] INSTALLATION I R,CASES I S. An improved shuffled frog leaping algorithm for assembly sequence planning of remote handling maintenance in radioactive environment[J]. Science&Technology of Nuclear Installations,2015(8):1376-1380.
[15] RAHIMI-VAHED A,DANGCHI M,RAFIEI H,et al. A novel hybrid multi-objective shuffled frog-leaping algorithm for a bicriteria permutation flow shop scheduling problem[J]. International Journal of Advanced Manufacturing Technology,2009,41(11-12):1227-1239.
[16] NAVEEN M,JAYARAMAN S,RAMANATH V,et al. Modified Levenberg-Marquardt algorithm for inverse problems[C]∥International Conference on Simulated Evolution and Learning. Springer-Verlag,2010:623-632.
[17] VASIN V V,PERESTORONINA G Y. The Levenberg-Marquardt method and its modified versions for solving nonlinear equations with application to the inverse gravimetry problem[J]. Proceedings of the Steklov Institute of Mathematics,2013,280(1):174-182.
[18]杨景豪,刘巍,刘阳,等.双目立体视觉测量系统的标定[J].光学精密工程,2016,24(2):300-308.YANG Jinghao,LIU Wei,LIU Yang,et al. Calibration of binocular vision measurement system[J]. Optics and Precision Engineering,2016,24(2):300-308.(in Chinese)
[19]李斌,王海峰,黄文倩,等.菠萝采收机械低成本双目视觉平台搭建与田间试验[J].农业工程学报,2012,28(增刊):188-192.LI Bin,WANG Haifeng,HUANG Wenqian,et al. Construction and in-field experiment of low-cost binocular vision platform for pineapple harvesting robot[J]. Transactions of the CSAE,2012,28(Supp.):188-192.(in Chinese)
[20] XIAO Z,JIN L,YU D,et al. A cross-target-based accurate calibration method of binocular stereo systems with large-scale field-of-view[J]. Measurement,2010,43(6):747-754.
[2] FHR G,JUNG C R. Camera self-calibration based on nonlinear optimization and applications in surveillance systems[J].IEEE Transactions on Circuits&Systems for Video Technology,2017,27(5):1132-1142.
[3] PITCHANDI N,SUBRAMANIAN S P. GA-based camera calibration for vision-assisted robotic assembly system[J]. Iet Computer Vision,2017,11(1):50-59.
[4] ZHANG Y,ZHOU L,LIU H,et al. A flexible online camera calibration using line segments[J]. Journal of Sensors,2016:Article ID 2802343.
[5]游江,唐力伟,邓士杰.基于改进遗传模拟退火的相机标定方法[J].计算机工程与设计,2017,38(3):819-824.YOU Jiang,TANG Liwei,DENG Shijie. Camera calibration method based on genetic algorithm and simulated annealing[J].Computer Engineering and Design,2017,38(3):819-824.(in Chinese)
[6]柯丰恺,陈幼平,谢经明,等.基于凸松弛优化算法的相机内外参数标定[J].中国机械工程,2016,27(5):634-638.KE Fengkai,CHEN Youping,XIE Jingming,et al. Internal and external parameter calibrations of a camera based on convex relaxation optimization algorithm[J]. China Mechanical Engineering,2016,27(5):634-638.(in Chinese)
[7]刘艳,李腾飞.对张正友相机标定法的改进研究[J].光学技术,2014,40(6):565-570.LIU Yan,LI Tengfei. Reaserch of the improvement of Zhang's camera calibration method[J]. Optical Technique,2014,40(6):565-570.(in Chinese)
[8] D'EMILIA G,GASBARRO D D. Review of techniques for 2D camera calibration suitable for industrial vision systems[C]∥Journal of Physics:Conference Series,2017,841:012030.
[9] ZHANG Z. A flexible new technique for camera calibration[J]. Tpami,2000,22(11):1330-1334.
[10] GARY B,ADRIAN K. Learning Open CV[M]. California:O'Reilly Media,2008.
[11] EUSUFF M,LANSEY K,PASHA F. Shuffled frog-leaping algorithm:a memetic meta-heuristic for discrete optimization[J].Engineering Optimization,2006,38(2):129-154.
[12] SARKHEYLI A,ZAIN A M,SHARIF S. The role of basic,modified and hybrid shuffled frog leaping algorithm on optimization problems:a review[J]. Soft Computing,2015,19(7):1-28.
[13] JADIDOLESLAM M,EBRAHIMI A. Reliability constrained generation expansion planning by a modified shuffled frog leaping algorithm[J]. International Journal of Electrical Power&Energy Systems,2015,64:743-751.
[14] INSTALLATION I R,CASES I S. An improved shuffled frog leaping algorithm for assembly sequence planning of remote handling maintenance in radioactive environment[J]. Science&Technology of Nuclear Installations,2015(8):1376-1380.
[15] RAHIMI-VAHED A,DANGCHI M,RAFIEI H,et al. A novel hybrid multi-objective shuffled frog-leaping algorithm for a bicriteria permutation flow shop scheduling problem[J]. International Journal of Advanced Manufacturing Technology,2009,41(11-12):1227-1239.
[16] NAVEEN M,JAYARAMAN S,RAMANATH V,et al. Modified Levenberg-Marquardt algorithm for inverse problems[C]∥International Conference on Simulated Evolution and Learning. Springer-Verlag,2010:623-632.
[17] VASIN V V,PERESTORONINA G Y. The Levenberg-Marquardt method and its modified versions for solving nonlinear equations with application to the inverse gravimetry problem[J]. Proceedings of the Steklov Institute of Mathematics,2013,280(1):174-182.
[18]杨景豪,刘巍,刘阳,等.双目立体视觉测量系统的标定[J].光学精密工程,2016,24(2):300-308.YANG Jinghao,LIU Wei,LIU Yang,et al. Calibration of binocular vision measurement system[J]. Optics and Precision Engineering,2016,24(2):300-308.(in Chinese)
[19]李斌,王海峰,黄文倩,等.菠萝采收机械低成本双目视觉平台搭建与田间试验[J].农业工程学报,2012,28(增刊):188-192.LI Bin,WANG Haifeng,HUANG Wenqian,et al. Construction and in-field experiment of low-cost binocular vision platform for pineapple harvesting robot[J]. Transactions of the CSAE,2012,28(Supp.):188-192.(in Chinese)
[20] XIAO Z,JIN L,YU D,et al. A cross-target-based accurate calibration method of binocular stereo systems with large-scale field-of-view[J]. Measurement,2010,43(6):747-754.
© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号 地址：北京市海淀区学院路29号 邮编：100083 电话：办公室：(+86 10)66554848；文献借阅、咨询服务、科技查新：66554700 |