| 融合线性方法和神经网络的摄像机并行标定技术 |
| |
| 引用本文： | 熊九龙,田震,夏军营,徐小泉.融合线性方法和神经网络的摄像机并行标定技术[J].仪器仪表学报,2011,32(6). |
| |
| 作者姓名： | 熊九龙 田震 夏军营 徐小泉 |
| |
| 作者单位： | 国防科技大学机电工程与自动化学院,长沙,410073 |
| |
| 摘 要： | 从摄像机成像模型的基础原理出发,针对神经网络的特性,构建不同的训练点集以获得不同的神经网络.提出对物点的世界坐标(XW,YW,ZW)信息分别进行标定,称之为并行标定.此方法不局限于单个平面位置的标定,标定过程中让标定模板沿ZW方向移动,覆盖一个三维标定区域.与以往标定方法相比,由于该方法减少了标定点数目,此方法在提高了标定精度的同时提高了标定速度.实验结果表明,该方法在XW轴和YW轴方向上的距离测量精度优于0.1%,在ZW轴方向上的距离测量精度优于1%.
|
| 关 键 词： | 摄像机标定 并行标定 神经网络 摄像机模型 |
Camera parallel calibration technology based on linear method and neural network |
| |
| Xiong Jiulong,Tian Zhen,Xia Junying,Xu Xiaoquan.Camera parallel calibration technology based on linear method and neural network[J].Chinese Journal of Scientific Instrument,2011,32(6). |
| |
| Authors: | Xiong Jiulong Tian Zhen Xia Junying Xu Xiaoquan |
| |
| Affiliation: | Xiong Jiulong,Tian Zhen,Xia Junying,Xu Xiaoquan(College of Mechatronic Engineering and Automation,National Univ.of Defense Technology,Changsha 410073,China) |
| |
| Abstract: | This paper discusses the fundamental principle of camera imaging model and the characteristic of neural network,constructs different training point sets to obtain different neural networks.The world coordinates(XW,YW,ZW) of the objective points are calibrated individually,which is called parallel calibration in this paper.This method can calibrate the model in a 3D area while the model is moving along ZW axis rather than only in a plane location.Compared with other calibration methods the presented method i... |
| |
| Keywords: | camera calibration parallel calibration neural network camera model |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
|
