复杂环境下高速多维运动目标位姿测量关键技术研究

This proposal focuses on the high-speed and multi-dimensional moving target pose measurement problem based on machine vision method under complex environment. The problem orients to the great demands in the aerospace, aviation, navigation and other vehicle fields for the precision control of high-speed and multi-dimensional moving target in order to make the core equipment work safely. In this proposal, the effect law of feature imaging with multi-constraint and complicated influence factor on small high-speed moving target in harsh environment is revealed, and the image acquisition method is presented to capture clear images which only contain the pose feature of small high-speed and multi-dimensional moving target. Meanwhile, based on the video synchronization and composite target, a method of camera field calibration with large visual field is obtained to ensure the high precision calibration. Then, by virtue of four dimensional time stream, a high-efficiency and high-precision way is also carried out to identify and match the feature of high-speed and multi-dimensional moving target, and to solve the feature points blanking and other problems which appear in the condition for high speed moving of the target in the relative large space. Furthermore, in order to improve the precision and robustness of the system against the changing conditions, researches are conducted to obtain a pose reconstruction method which is based on prior model and a sensitivity relational model of the total uncertainty of multi-dimensional moving target measuring and the local measurement error. Finally, the measurement system for high-speed and multi-dimensional moving target is established, and experiments are conducted to verify and improve the stability and practicability of this measuring method. The results may provide significant testing technology support for the safety and efficiency of working equipment in complicated working conditions.

面向航天、航空、航海、运载等领域中核心作业装备安全运行对高速多维运动目标精确控制的迫切需求，研究基于机器视觉法的复杂环境下高速多维运动目标位姿测量难题，揭示恶劣服役环境中多约束条件与复杂影响因素对小型高速运动目标特征成像的影响规律，提出仅蕴含位姿特征的小型高速多维运动目标图像采集方法，以获取目标清晰图像；研究基于视频同步的复合靶标的大视场摄像机标定方法，实现摄像机现场高精度标定；建立基于四维时间流的高速多维运动目标特征高效精准匹配方法，以解决小型目标在较大空间内高速运动时特征点消隐等问题；研究基于先验模型的高速多维运动目标位姿还原方法，建立多维运动目标位姿总体测量不确定度与测量环节误差的敏度分析模型，以提高测量条件变化时系统的精度和鲁棒性；构建高速多维运动目标位姿测量系统，验证并完善目标位姿测量方法的可靠性和实用性。研究成果可为复杂服役环境下机械作业装备安全高效运行提供实验和技术支撑。

航空、航天等领域高端装备运行过程中往往需对其高速运动部件的位置和姿态进行高精度测量，以确保系统安全可靠。而这类高速运动目标的多维位姿测量受强流场、低照度等复杂环境影响严重，且存在高动态测量、大量程内小目标高精测量、多维位姿高可靠测量等特点，导致其测量难度大。视觉测量技术虽优势明显，但易受复杂环境影响，导致精度低、环境适应性差。因此，本项目开展复杂环境下高速多维运动目标位姿的视觉测量技术的应用基础研究，以期提升我国复杂服役环境下系统运动参数动态测量技术水平。. 本项目以超高速飞行器风洞投放试验目标为研究对象，突破了清晰图像采集、微弱特征精确提取、位姿可靠估算等目标位姿视觉测量瓶颈难题，形成了一套复杂环境下高速多维运动目标位姿视觉测量关键技术。针对低照度、强流场风洞环境导致的目标图像质量差、特征模糊的问题，提出了低照度环境下蕴含目标多维位姿信息的高信噪比序列图像采集方法，获得的目标图像清晰、特征明显；针对风洞测量中视场大、光路介质折射率多变等导致成像畸变大、精度低的问题，提出了复杂环境下目标多折射成像模型，建立了基于分步标定的大视场摄像机标定方法，实现了测量系统高精度标定；针对序列图像特征信息少、干扰大等导致的目标特征还原精度低的问题，提出了基于时空约束的高速运动目标序列图像处理与多维位姿还原方法，实现了多维位姿快速高精度求取；研发出高速目标投放位姿测量系统，开展超音速风洞下高速多维运动目标位姿测量实验研究，小型风洞模型位置测量精度优于0.103mm，俯仰和偏航角度精度优于0.107°，滚转角度精度优于0.648°；大滚转角类模型位置测量精度优于0.07mm，俯仰和偏航角度精度优于0.077°，滚转角度精度优于0.207°，解决了系列超高速风洞模型的投放位姿测量难题，并在中航工业沈阳空气动力研究院获得成功应用。本项目研究的复杂环境下高速多维运动目标位姿的视觉测量关键技术成果亦对其它复杂环境下装备与系统状态参量精密测量具有参考价值。