一种双目立体视觉系统的校准方法
|
摘要
基于双目立体视觉系统的图像分析以及人工神经网络的三维空间建模算法,设计了一种针对双目立体视觉相机的校准方法,并可应用于运动目标点的轨迹追踪。将均匀分布目标点的校准平面放置在有效视野内的不同位置,通过双目立体视觉系统来捕获处于不同位置的校准平面图像。在图像处理之后,使用校准点中心的二维坐标作为人工神经网络训练的输入样本集,通过建立人工神经网络模型结构,实现目标点二维平面坐标到三维空间坐标的映射关系。采用这种具有通用性的方法,可以有效修正系统中存在的失真因子,获得目标三维位置信息,而无需进行复杂的相机校准操作。实验表明,提出的方案具有良好的可行性和鲁棒性。
The camera calibration method of binocular vision system which can be used in trajectory tracking is based on image analysis and three-dimensional spatial position modeling algorithm of artificial neural network. The calibration plane with uniformly distributed target points was placed in multiple positions within the camera field. The images of calibration plane are captured by the binocular vision system. Afer image processing,the coordinates of the target points are determined. The coordinates of the points are used as input data set of the artificial neural network. Througth optimizing parameters of the artificial neural network,the mapping relationship between the two-dimensional coordinates of the target point and the three-dimensional spatial coordinates is determined. With this versatile method,distortion factors of the binocular vision camera system can be eliminated and three-dimensional position information without complicated proceduring of camera calibration operation. The experiment result demonstrates that the calibration method has good feasibility and robustness.
The camera calibration method of binocular vision system which can be used in trajectory tracking is based on image analysis and three-dimensional spatial position modeling algorithm of artificial neural network. The calibration plane with uniformly distributed target points was placed in multiple positions within the camera field. The images of calibration plane are captured by the binocular vision system. Afer image processing,the coordinates of the target points are determined. The coordinates of the points are used as input data set of the artificial neural network. Througth optimizing parameters of the artificial neural network,the mapping relationship between the two-dimensional coordinates of the target point and the three-dimensional spatial coordinates is determined. With this versatile method,distortion factors of the binocular vision camera system can be eliminated and three-dimensional position information without complicated proceduring of camera calibration operation. The experiment result demonstrates that the calibration method has good feasibility and robustness.
引文
[1]Hu X C,An H L,Ma H X.Genetic optimized BP network method for camera calibration in binocular vision[J].Trans Tech Publ,2013,756(1):3404-3409.
[2]袁野,欧宗瑛,田中旭.应用神经网络隐式视觉模型进行立体视觉的三维重建[J].计算机辅助设计与图形学学报,2003,15(3):293-296.Yuan Y,Ou Z Y,Tian Z X.3D reconstruction of stereo vision using neural networks implicit vision model[J].Journal of Computer Aided Design and Computer Graphics,2003,15(3):293-296.
[3]韦争亮,古耀达,黄志斌,等.双目立体视觉中特征点三维坐标重构校准研究[J].计量学报,2014,35(2):102-107.Wei Z L,Gu Y D,Huang Z B,et al.Research on Calibration of Three Dimensional Coordinate Reconstruction of Feature Points in Binocular Stereo Vision[J].Acta Metrologica Sinica,2014,35(2):102-107.
[4]Jia Z,Yang J,Liu W.Improved camera calibration method based on perpendicularity compensation for binocular stereo vision measurement system[J].Optics express,Optical Society of America,2015,23(12):15205-15223.
[5]Turton B C,Arslan T,Horrocks D H.A hardware architecture for a parallel genetic algorithm for image registration[J].The Institution of Engineering and Technology,1994,1(11):1-6.
[6]Chu F,Wang L.Applications of support vector machines to cancer classification with microarray data[J].International journal of neural systems,2005,15(6):475-484.
[7]Fu X,Wang L.Data dimensionality reduction with application to simplifying RBF network structure and improving classification performance[J].IEEE Transactions on Systems,Man,and Cybernetics,Part B(Cybernetics),2003,33(3):399-409.
[8]Zhang K,Gao Z.The Comparison of Calibration Method of Binocular Stereo Vision System[C]//3rd International Conference on Material,Mechanical and Manufacturing Engineering,2015,1(1):894-901.
[9]Zhao Y,Gong L,Huang Y.A review of key techniques of vision-based control for harvesting robot[J].Computers and Electronics in Agriculture,2016,127:311-323.
[10]陈智军,韩超,陈涛,等.基于人工神经网络的乐甫波液体密度粘度并行检测研究[J].计量学报,2017,38(6):721-724.Chen Z J,Han C,Chen T,et al.Research on Parallel Measurement of Liquid Density and Viscosity Using Love Wave Based on Artificial Neural Networks[J].Acta Metrologica Sinica,2017,38(6):721-724.
[2]袁野,欧宗瑛,田中旭.应用神经网络隐式视觉模型进行立体视觉的三维重建[J].计算机辅助设计与图形学学报,2003,15(3):293-296.Yuan Y,Ou Z Y,Tian Z X.3D reconstruction of stereo vision using neural networks implicit vision model[J].Journal of Computer Aided Design and Computer Graphics,2003,15(3):293-296.
[3]韦争亮,古耀达,黄志斌,等.双目立体视觉中特征点三维坐标重构校准研究[J].计量学报,2014,35(2):102-107.Wei Z L,Gu Y D,Huang Z B,et al.Research on Calibration of Three Dimensional Coordinate Reconstruction of Feature Points in Binocular Stereo Vision[J].Acta Metrologica Sinica,2014,35(2):102-107.
[4]Jia Z,Yang J,Liu W.Improved camera calibration method based on perpendicularity compensation for binocular stereo vision measurement system[J].Optics express,Optical Society of America,2015,23(12):15205-15223.
[5]Turton B C,Arslan T,Horrocks D H.A hardware architecture for a parallel genetic algorithm for image registration[J].The Institution of Engineering and Technology,1994,1(11):1-6.
[6]Chu F,Wang L.Applications of support vector machines to cancer classification with microarray data[J].International journal of neural systems,2005,15(6):475-484.
[7]Fu X,Wang L.Data dimensionality reduction with application to simplifying RBF network structure and improving classification performance[J].IEEE Transactions on Systems,Man,and Cybernetics,Part B(Cybernetics),2003,33(3):399-409.
[8]Zhang K,Gao Z.The Comparison of Calibration Method of Binocular Stereo Vision System[C]//3rd International Conference on Material,Mechanical and Manufacturing Engineering,2015,1(1):894-901.
[9]Zhao Y,Gong L,Huang Y.A review of key techniques of vision-based control for harvesting robot[J].Computers and Electronics in Agriculture,2016,127:311-323.
[10]陈智军,韩超,陈涛,等.基于人工神经网络的乐甫波液体密度粘度并行检测研究[J].计量学报,2017,38(6):721-724.Chen Z J,Han C,Chen T,et al.Research on Parallel Measurement of Liquid Density and Viscosity Using Love Wave Based on Artificial Neural Networks[J].Acta Metrologica Sinica,2017,38(6):721-724.