摘要
在单目视觉检测中,由于相机设置角度以及镜头等原因,相机所拍摄的图片存在非线性畸变,在图片使用之前通常需要进行相机标定、畸变校正。径向基函数(RBF)神经网络具有良好的非线性拟合能力,文中提出一种基于RBF神经网络的相机标定方法,该方法从标准标定板中提取样本像素点作为网络输入,对神经网络进行训练,训练好的网络可以得到畸变图像与校正图像像素点之间的对应关系,进而达到校正图像畸变的作用。实验结果表明,相较于传统的张正友标定法,该方法简单有效,精确度更高。
In monocular vision inspection,due to the setting angle of the camera and the lens,the pictures taken by the camera have non-linear distortions,and the camera calibration is usually required for distortion correction before the pictures are used.Radial basis function(RBF) neural network has good nonlinear fitting ability.A camera calibration method based on RBF neural network is proposed.This method extracts sample pixels from the standard calibration board as the network input to train the neural network.The trained network can obtain the corresponding relationship between the distorted image and the pixel points of the corrected image,and then achieve the function of correcting the image distortion.The experimental results show that compared with the traditional Zhang Zhengyou calibration method,this method is simple and effective,and has higher accuracy.
作者
刘驰弋
谢明红
何浪
LIU Chiyi;XIE Minghong;HE Lang(College of Mechatronics and Automation,Huaqiao University,Xiamen 361021,China)
出处
《机械工程师》
2022年第7期58-61,共4页
Mechanical Engineer
关键词
单目视觉
相机标定
图像校正
RBF神经网络
monocular vision
camera calibration
image correction
RBF neural network
作者简介
刘驰弋(1997—),男,硕士,研究方向为计算机视觉、计算机图形学;通信作者:谢明红(1968—),男,博士,研究员,主要研究方向为数控技术、计算机视觉,172399251@qq.com;何浪(1996—),男,硕士,研究方向为计算机视觉、数控技术。
