Abstract In view of object pose measurement based on monocular vision, the relationship between the input parameters errors of a camera and the measurement error of pose was studied. Three types of input parameters errors were discussed theoretically, and an isosceles triangle Perspective-3-Point model (P3P model) was built, composed by three feature points. Thus the relationship between input parameters error of camera and pose measurement error was finally derived. Adding types of noise to every input parameter error during the experiment, observing how each of the input parameters errors influences the pose measurement error. Experiment results reveal the absolute value of correlation coefficients between post measurement error and three types of input parameter errors is superior to 0.99, within the margin of error, suggesting approximately linear relationship among them. Furthermore, the attitude angle error as well as the position error caused by cameras intrinsic parameters and 2D image coordinates are 10 and 100 times respectively of that caused by 3D model coordinates.
YAO Zhen-jian,WANG Zhong-yu,LI Dan. Error Analysis of Pose Measurement with Monocular Vision Based on Isosceles Triangle Model[J]. Acta Metrologica Sinica, 2014, 35(z1): 19-25.
［1］黄桂平,李广云,王宝丰. 单目视觉测量技术［J］. 计量学报, 2004, 25(4): 314-317.
［2］Yi Y M, Liu D. Simultaneous localization and mapping using monocular vision［J］. Journal of Nanoelectronics and Optoelectronics, 2012, 7(2): 162-166.
［3］韩延祥,张志胜,戴敏. 用于目标测距的单目视觉测量方法［J］. 光学精密工程, 2011, 19(5): 1110-1117.
［4］余厚云,张为公. 基于单目视觉的跟驰车辆车距测量方法［J］. 东南大学学报（自然科学版）, 2012, 42(3): 542-546.
［5］Yu D, Wei W, Zhang Y H. Person following of mobile robot using monocular camera ［J］. Chinese Journal of Scientific Instrument, 2010, 31(3): 659-664.
［6］吴凤和. 基于计算机视觉测量技术的图像轮廓提取方法研究［J］. 计量学报, 2007, 28(1): 18-22.
［7］Joaquim Salvi, Xavier Armangue, Joan Batlle. A comparative review of camera calibrating methods with accuracy evaluation［J］. Pattern Recognition, 2002 (35): 1617-1635.
［8］余厚云, 张为公. 基于单目视觉传感器的车距测量与误差分析［J］. 传感器与微系统, 2012, 31(9): 10-13.
［9］弋英民, 刘丁. 基于单目视觉的轮式机器人同步定位与地图构建［J］. 仪器仪表学报, 2010,31(1): 117-121.
［10］Kim Minho, Sihyun Joo,Sungho Jo. 3D human-pose tracking through a monocular vision［J］. Advances in Intelligent Systems and Computing, 2013,208: 241-253.
［11］Fishler M A, Bolles R C. Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography［J］. Communication of the ACM, 1981, 24(6): 381-395.
［12］Wang J, Qiu Z R, Li P Y. Application of parameter's P3P model in vision coordinates measurement process［J］. Infrared and Laser Engineering, 2010, 39(1): 151-155.
［13］孙凤梅,王波. 关于PnP问题多解的分布与解的稳定性的讨论［J］. 自动化学报, 2010, 36(9): 1213-1219.
［14］周鑫,朱枫. 关于P3P问题解的唯一性条件的几点讨论［J］. 计算机学报, 2003, 26(12): 1696-1701.
［15］Kamgar-Parsi B. Evaluation of quantization error in computer vision［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1989, 11(9): 929-940.
［16］郝颖明,朱枫,欧锦军,等. P3P位姿测量方法的误差分析［J］. 计算机工程与应用, 2008, 44(18): 239-242.
［17］周静,朱枫,郝颖明,等. 摄像机标定参数误差对位姿测量精度的影响［C］ //中国仪器仪表学会第九届青年学术会议,安徽黄山, 2007.
［18］王中宇,刘智敏,夏新涛,等. 测量误差与不确定度评定［M］. 北京:科学出版社,2008.
［19］Jian R, Kasturri R, Schunck B G. Machine Vision［M］. Beijing: China Machine Press, 2003.
［20］李定坤,叶声华,任永杰. 机器人定位精度标定技术的研究［J］. 计量学报, 2007, 28(3): 224-227.
［21］王中宇,葛乐矣,杨文平,等. 一种小样本虚拟仪器测量不确定度评定新方法［J］. 计量学报, 2008, 29(4): 387-392.
［22］李慧. 光电经纬仪测量精度分析的误差模型研究［D］. 长春:中国科学院长春光学精密机械与物理研究所, 2008.
2014, Vol. 35 
