作者：游江, 唐力伟, 邓士杰

作者单位：军械工程学院火炮工程系, 河北 石家庄 050003

关键词：机器视觉;自动分割;自动排序;向量夹角;姿态测试

摘要：

针对空间目标姿态测量问题，提出基于机器视觉的空间目标姿态自动测量方法，并对其中较为关键的目标靶自动分割及特征点自动排序问题进行研究。首先改进圆形目标靶的设计，通过设置区域面积及矩形度阈值，实现目标靶的自动分割，并提出基于向量夹角的圆心点阵顺序排序法，实现特征点准确快速地自动排序，最终基于张正友标定法，完成相机标定，获取目标靶在相机坐标系下的旋转矩阵，进而计算目标靶的空间姿态变化。实验结果验证：姿态解算的最大相对误差3%，整个姿态解算过程自动化程度较高，平均耗时2.026 s，能够满足工程测试需要。

Research of automatic measurement method of space target's posturebased on machine vision

YOU Jiang, TANG Liwei, DENG Shijie

Department of Guns Engineering, Ordnance Engineering College, Shijiazhuang 050003, China

Abstract: Aiming at the problem of the measurement of target's posture,a non-contact automatic measurement system of space posture based on the machine vision was established,and the key steps of automatic division of the target and ranking of the points have been researched.Firstly the design of target was improved,and then the automatic division of the target was realized by setting the area and rectangularity threshold.Then the ranking arithmetic based on vector angle was proposed to realize the fast and exactly automatic ranking of the points.Finally the camera was calibrated by Zhang's method,and the conversion of the target's posture was calculated.The experiment result shows that:the largest relative error of the measurement of the posture is no more than 3%,and the automatic testing method is of high automaticity,and the average time of posture testing is about 2.026 s.Therefore the proposed method meets the engineering test requirement.

Keywords: machine vision;automatic division;automatic ranking;vector angle;posture measurement

2016, 42(11): 107-112  收稿日期: 2016-5-10;收到修改稿日期: 2016-5-10

基金项目: 

作者简介: 游江(1992-),男,河南南阳市人,硕士,主要从事机器视觉及性能测试方面研究。

参考文献

[1] 韩冬,何闻.超低频标准振动系统基础设计技术[J].振动与冲击,2008,27(9):20-22。
[2] 滑广军,李强,杨益民,等.某大型超静定框架的动态应力测试与有限元分析[J].振动与冲击,2008,27(8):170-172
[3] CHUNG H,LIANG J,KUSHIYAMA S,et al.Digital image processing for non-linear system identification[J].International Journal of Non-linear Mechanics,2004,39(5):691-707.
[4] 歹英杰,汪伟,邓士杰,等.基于立体机器视觉的动目标空间位姿测试研究[J].振动与冲击,2015,34(16):188-212.
[5] 朱晓林,高诚辉,何炳蔚,等.基于机器视觉的机械零件图像尺寸自动标注方法[J].中国工程机械学报,2010,8(3):324-328.
[6] 张梁,徐锦法,夏青元.双目立体视觉的无人机位姿估计算法及验证[J].哈尔滨工业大学学报,2014,46(5):66-72.
[7] 陈爱华,高诚辉,何炳蔚.立体视觉中的图像立体匹配算法[J].中国工程机械学报,2014,12(3):194-198.
[8] 郭继平,李阿蒙,于冀平,等.双目立体视觉动态角度测量方法[J].中国测试,2015,41(7):21-23,36.
[9] SHI X F,HONG N,CAI T F.An automatic and robust chessboard corner extraction[J].Transaction on Edutainment VI,2011(6758):49-56.
[10] 夏仁波,刘伟军,赵吉宾,等.基于圆形标志点的全自动相机标定方法[J].仪器仪表学报,2009,30(2):368-373.
[11] 徐鹏,汪建业,王燕儒.摄像机标定中靶标圆心像点坐标的精确计算[J].红外与激光工程,2011,40(7):1342-1346.
[12] 张春燕,邹伟.一种鱼眼镜头标定板的设计、检测与排序方法[J].计算机工程与应用,2015,51(15):188-192.
[13] 梁力,尹东斐,王川.高精度摄像机标定模板的设计及设别算法[J].西安交通大学学报,2011,45(4):82-85.
[14] BOUGUET J.Visual methods for three-dimensional modeling[D].Pasadena:California Institute of Technology,1999.
[15] 胡茂林.空间和变换[M].北京:科学出版社,2007:123-124.