摘要
四轮定位参数对于汽车的操控性、安全性以及耗油量等有重大影响。在阐述汽车四轮前束和外倾、主销内倾和后倾等主要定位参数以及3D四轮定位仪结构和测量原理的基础上,针对机器视觉3D四轮定位仪开发过程中的关键技术,基于图像特征点的灰度模型进行棋盘格型角点的亚像素精度提取;基于平面靶标进行相机内部参数的标定;采用旋转向量一致的双平面靶标进行相机外部参数的标定;基于靶标旋转平移过程中的几何关系,采用最小二乘法优化计算轮胎中心点和旋转轴向量。采用1 600像素×1 200像素、像素尺寸3.0μm×3.0μm、USBX.0接口工业相机,940 nm波长LED环形光源,8 mm焦距低畸变工业相机镜头,6×6棋盘格型靶标开发出商品级的3D四轮定位仪,并与硬件档次相当的某国际品牌四轮定位仪进行了对比测试,基本达到设计精度。该商品已投放市场,用户反映良好。
Four-wheel alignment parameters have important effects on the vehicle hamlling capacity,safety perforul- ance and fuel consumption. The main alignment parameters, such as toe angle,camber angle,kingpin inclination and caster angles are explained. The structure and measurement principle of 3D computer vision four-wheel aligner are described. The key technologies of 3D four-wheel aligner based on machine vision were developed, including the sub- pixel positioning technique for extracting the x-corners of chessboard using partial grey-level model of the image fea- tures ,the calibration technique for camera interior parameters using plane target ,the calibration technique for camera exterior parameters using two plane targets with invariable pose and position relation, and the optimization method of least squares for detecting the rotation centers and axis vectors of 4 wheels based on certain geometric motion relations in target rotation and translation. A commercial 3D four-wheel aligner was developed, employing USB3.0 industrial cameras of 1 600 x 1 200 pixels with the pixel dimension of 3.0 Ixm x 3.0 ~m, LED lighting sourees with the wave- length of 940 nm,8 mm focus low-distortion industrial camera lens ,and a 6 ~ 6 chessboard target. The designed four- wheel aligner and a certain international brand four-wheel aligner in the same hardware rank were tested and com- pared. The test results show that the developed product has reached the design accuracy, and the products have launched to the market with excellent consumer responses.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2013年第10期2184-2190,共7页
Chinese Journal of Scientific Instrument
基金
国家自然科学基金(51275169
50875083)
湖南省自然科学基金(11JJ9016)
湖南省高校科技创新团队资助项目
关键词
四轮定位仪
相机标定
角点提取
机器视觉
four-wheel aligner
camera calibration
corner extraction
computer vision
