【作者】 王习文；

【导师】 陈娟；

【作者基本信息】 中国科学院研究生院（长春光学精密机械与物理研究所） ， 机械电子工程， 2010， 博士

【摘要】 光电经纬仪获取的跟踪测量图像,记录了目标飞机运动过程的各种事件和姿态信息:如空间位置、运动轨迹、运动速度、运动加速度及俯仰、偏航、滚动的3D姿态变化等。但传统光电跟踪图像处理将目标飞机等效为一个质点,因而得不到飞机的3D姿态信息,致使图像中蕴含的目标飞机外形形状及姿态信息都舍弃了。本文开展的光电经纬仪跟踪飞机的3D计算机视觉研究,就是设法利用光电经纬仪获得的跟踪数据,去充分挖掘跟踪图像中隐含的信息:既得到飞机的运动信息,又得到飞机的3D姿态信息,并可在事后还原飞机当时的运动情形,以便对飞机的飞行性能进行分析、判断、评估和验证,或者在试验分析、事故鉴定、训练评估、飞控系统检验及新型号的设计定型等方面发挥相应的作用。为此本论文主要围绕三个方面的问题展开:一是依据光电经纬仪的跟踪数据,求解出飞机在跟踪测量期间的运动轨迹和运动参数;二是依据跟踪图像和跟踪数据,求解出飞机在跟踪测量期间的3D姿态;三是建立相应的仿真模型,按跟踪测量的时间节点,关联飞机的运动轨迹、运动参数和3D姿态参数,在计算机仿真环境中驱动事先建好的飞机3D模型,还原飞机在跟踪测量期间的运动全过程。首先根据光电经纬仪的测量原理,推导出单站跟踪,双站交会跟踪两种模式下确定目标空间位置的基本公式,据此给出光电经纬仪测量坐标系、图像坐标系、地心坐标系、发射坐标系等空间坐标系间的转换关系;其次以形心(或质心)跟踪测量方法,确定目标飞机在空间的定位,用多项式曲线拟合方法得到其在特定坐标系X、Y、Z三个方向的运动轨迹方程,对轨迹方程求取一阶、二阶导数,给出相应的速度、加速度方程,并对运动轨迹、运动速度和加速度的求解进行精度分析;然后在现有3D姿态测量方法的基础上,提出测量飞机3D姿态的角平分线方向向量法,利用2台经纬仪交会测量得到的飞机图像序列,通过图像预处理、边缘检测、特征信息提取、机翼边缘特征线方向向量求解,机翼角平分线的方向向量求解等过程,得到各跟踪测量时间节点时飞机的俯仰、偏航和滚动3D姿态,并对计算结果进行相应的精度分析;再利用三维CAD软件建立目标飞机外形的3D几何模型,导入到3DSmax软件中,关联上运动轨迹、运动参数和3D姿态信息,建立了光电经纬仪跟踪测量飞机的仿真模型,成功地模拟光了电经纬仪跟踪测量飞机的运动情况,获取以跟踪时间为序的单站和双站交会跟踪的图像序列;最后通过分析飞机的眼镜蛇机动过程,对本文提出的光电经纬仪跟踪飞机的3D计算机视觉过程进行仿真实验验证,在计算机仿真环境中再现了飞机的运动和姿态随跟踪时间的变化情况,结果表明所提出的方法合理、可行,具有实用性。更多还原

【Abstract】 The measuring images of aircraft tracked by photoelectric theodolite, record the various events and attitudes information of the aircraft during the motion process: such as space position, moving trajectory, moving speed, moving acceleration and pitch, yaw, roll changes. However, make the aircraft as a particle in traditional image process of electro-optical tracking, can not obtain the 3D attitudes information, so the shape and attitude information of the aircraft is discarded. The study of 3D computer vision for photoelectric theodolite tracking aircraft carried out in this dissertation, is to fully acquire the hidden information exist in the image by the optical theodolite tracking data, not only the aircraft movement information, but also the aircraft’s 3D attitudes. Base on that, the aircraft’s flight performance can be analyzed, judged, evaluated and validated, or it plays a corresponding role in test analysis, incident identification, training evaluation, flight control system testing and new model’s designing.In order to achieve those purposes, this dissertation mainly includes three aspects: The first is to solve the moving trajectory and motion parameters of aircraft during the process of being tracked and measured, based on photoelectric theodolite tracking data; The second is to solve the aircraft’s 3D attitudes based on the tracking images and data; The third is to establish the corresponding simulation model, according to tracking and measuring time nodes associated with the aircraft moving trajectory, motion parameters and 3D-attitude parameters, to drive the prebuilt aircraft’s 3D model run in computer simulation environment, and restore the entire process of the aircraft being tracked and measured. First of all, under the condition of single-theodolite tracking and two-theodolite intersection tracking, the basic formula of the aircraft’s space location are built, according to the measurement principle of photoelectric theodolite. Depending on that, the transfer relations among the photoelectric theodolite coordinate measurement system, the image coordinate system, the geocentric coordinate system and the launch coordinate system is given. Then dentermine the aircraft’s space position by shape center(or weight center) tracking measurement, and using polynomial curve fitting method to obtain the moving trajectory equations along X, Y, Z three directions in a certain coordinate system. The first-order derivative equations of the trajectory equations are exactly the velocity equations, and the second-order derivative equations of the trajectory equations are exactly the acceleration equations. The accuracy analysis of moving trajectory, velocity and acceleration equations are carried out. After that, put forward the direction vector method of angle bisector lines for measuring the aircraft’s 3D attitudes on the basis of exist methods. The aircraft’s measuring images intersection tracked by two theodolites, through the process of image preprocessing, edge detection, feature information extraction, the direction of the wing edge characteristic line vector solving, direction vectors of two angle bisector lines solving, the pitch, yaw and roll of aircraft’s 3D attitudes at all the tracking time nodes will be acquired, and the precision of 3D attitude’s calculation results is analyzed. Afterwards, the 3D model of target aircraft is built by 3D CAD software, import the aircraft’s 3D model into 3DSmax software, associated it with the moving trajectory, moving parameters and 3D attitudes to establish the simulation model of photoelectric theodolite tracking aircraft. Using it simulate successfully the process of the electrical optical theodolite tracking the aircraft’s movement, and obtain the series images of aircraft’s tracked by single-theodolite and intersection tracked by two-theodolite. At last, example for the aircraft’s cobra maneuver, to verify the solution of 3D computer vision for photoelectric theodolite tracking aircraft which proposed in the dissertation reproduce the aircraft movement and 3D attitudes changes over the tracking time, the results indicate that the above-mentioned method is reasonable, feasible and practical.更多还原