基于神经网络的振镜式线结构光三维测量系统的标定
2019-10-12
作者：杨林林1，杨树明2
单位：1中国工程物理研究院机械制造工艺研究所；2西安交通大学机械制造系统国家重点实验室
摘要：振镜式线结构光三维测量系统是一种新型系统，具有扫描速度快、测量精度和集成度高等优点，在三维测量领域具有广泛的应用前景。为了避免建立复杂的系统数学模型，提高系统的标定精度，提出了一种基于神经网络的标定方法。该方法首先基于双重交比不变原理，利用棋盘格靶标和电动平移台获取大量精确的标定点坐标，以标定点的图像坐标及光平面转角为输入，对应的三维世界坐标为输出，建立系统的神经网络模型，通过训练完成系统标定。试验表明，本文提出的标定方法简化了系统的标定过程，标定精度高且通用性好。
关键词：振镜；线结构光；神经网络；标定方法
中图分类号：TG87；TH741；TP183文献标志码：ADOI：10.3969/j.issn.1000-7008.2019.10.022
Calibration of Galvanometric Linestructured Light Based on Neural Network
Yang Linlin，Yang Shuming
Abstract:The threedimensional measurement based on galvanometric linestructured light is a new system，which has widly application prospects in the field of threedimensional measurement due to its advantages of fast scanning speed，high measurement accuracy and high integration.In order to avoid establishing the complex mathematical model and improve the calibration accuracy of the system，a calibration method based on neural network is proposed in this paper.Based on the principle of double cross ratio invariance，a large number of precise coordinates of the calibration points are obtained by using chessboard target and electric translationplatform.By using the image coordinates of calibration points and the rotation angle of laser plane as input，and the corresponding threedimensional world coordinates as output，the neural network model of the system is established and the system calibration is completed by training.The experiments show that the proposed calibration method simplifies the calibration process and has high calibration accuracy and good versatility.
Keywords:galvanometric；linestructured light；neural network；calibration