基于高斯过程回归的图像角度传感器补偿技术  

Image angle sensor compensation technology based on Gaussian process regression

作　　者：杨德升 万璞 张靖 李鑫 刘音序 王凤 YANG Desheng;WAN Pu;ZHANG Jing;LI Xin;LIU Yinxu;WANG Feng(School of Physics and Information Engineering,Zhaotong University,Zhaotong 657000,China)

机构地区：[1]昭通学院物理与信息工程学院,云南昭通657000

出　　处：《传感器与微系统》2025年第3期58-61,66,共5页Transducer and Microsystem Technologies

基　　金：云南省科技计划项目(202303AA080041);定向人才培养训练项目(20230111453)。

摘　　要：为了解决现有光电编码器结构复杂、成本较高的问题,设计了一种基于图像识别光斑中心点的角度传感器。随着旋转轴转动的激光发射器发出的激光在底板上形成一个光斑,利用摄像头采集这个光斑的图像,通过高斯拟合法计算出光斑的中心点坐标,进而转换成对应的旋转角度。利用多项式拟合算法对采集到的角度—像素坐标之和曲线进行拟合分析,并采用高斯过程回归算法对传感器存在的非线性特性进行补偿。实验结果表明:采用高斯过程回归算法进行补偿后,传感器的非线性误差从7.745降低至2.840,测量精度达到了0.79°,其迟滞性、灵敏度和分辨率分别为±2.71%,2.502像素/(°)和2735像素/(°)。In order to solve the problem of complex structure and high cost of existing photoelectric encoders,an angle sensor based on image recognition spot center is designed.The laser emitted by the laser emitter rotating with the rotation axis forms a light spot on the bottom plate.The image of this light spot is collected by the camera,and the coordinates of the center point of the light spot are calculated by Gaussian fitting method,and then converted into corresponding rotation angle.Polynomial fitting algorithm is used to fit and analyze the collected angular-pixel coordinate sum curve,and Gaussian process regression algorithm is used to compensate the nonlinear characteristics of the sensor.The experimental results show that after compensated by Gaussian process regression algorithm,the nonlinear error of the sensor is reduced from 7.745 to 2.840,and the measurement precision reaches 0.79.Its hysteresis,sensitivity,resolution ratio are±2.71%,2.502pixel/(°),2735pixel/(°),respectively.

关 键 词：图像角度传感器 高斯拟合法 多项式拟合 高斯过程回归 非线性补偿 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置] O439[自动化与计算机技术—控制科学与工程]