POS辅助空中三角测量中的定权研究

发布时间：2018-04-23 04:19

本文选题：POS + 空中三角测量　；参考：《西南交通大学》2013年硕士论文

【摘要】：长期以来,航空摄影测量中影像的定向都需要借助于地面控制点进行传统的空中三角测量。近年来,随着传感器、空间定位技术以及计算机技术的迅速发展,非摄影测量信息正逐步取代传统的地面控制,在航空摄影测量中的应用得到快速发展。POS辅助空中三角测量作为POS与航空摄影测量最有效的集成方式,逐步成为航空摄影测量的研究热点和今后发展的趋势,如何提高其定位精度则成为研究的关键。与传统的空三加密方法相比,POS辅助空中三角测量在一定程度上可以减少甚至免去地面控制,这样既降低了生产成本,缩短了成图周期,同时提高了生产效率。然而,在研究POS系统稳定度和可靠性的同时,如何保证POS辅助空中三角测量的定位精度则显得尤为重要。目前对该问题的专注度主要集中在系统误差的影响上,往往是通过改变平差函数模型达到误差补偿的目的。但是,POS辅助空中三角测量是非摄影测量观测值与摄影测量观测值进行联合平差,参与平差的观测值不再是单一的像点坐标,还包括GPS(全球定位系统)所测得的线元素和IMU(惯性测量单元)所测得的角元素。所以如何合理地确定各种类型观测值的权,将直接影响平差结果的质量。因此,面对观测值类型的多样化,需要对POS辅助空三的随机模型进行正确合理的确定。本文针对经系统误差校正后的POS系统观测值和摄影测量像点坐标观测值,分别采用传统定权方法和验后方差分量估计方法,基于.NET平台编制了POS数据辅助光束法空三联合平差程序,并在某测区开展了实验研究,重点分析了不同的定权方法对平差结果精度的影响规律。实验结果表明,采用传统定权方法中的单位权定权和以常规的量测精度定权,加密点精度差异不大；采用验后方差分量估计方法进行平差,解算的加密点精度比传统定权方法有大幅度的提高；不过,采用同一验后方差分量估计方法,但取不同的初始权时会影响迭代计算的收敛速度；采用3种不同验后估计方法时,加密点精度稍有差异,但因Helmert估计始终无偏,而Fostner估计和Ebner估计过程均为从有偏至无偏,导致迭代次数有所差别,Helmert估计、Fostner估计和Ebner估计分别迭代了8次、11次和15次；同时,在平差过程中无论采用哪种定权方法,得到的加密点的Z向精度始终比X、Y方向的精度低。实验验证了验后方差分量估计在POS辅助光束法区域网空中三角测量中的必要性和有效性。
[Abstract]:For a long time, the orientation of aerial photogrammetry requires the help of ground control points to carry out traditional aerial triangulation. In recent years, with the rapid development of sensors, space positioning technology and computer technology, non-photogrammetric information is gradually replacing the traditional ground control. As the most effective integration method between POS and aerial photogrammetry, the application in aerial photogrammetry has gradually become the research hotspot of aerial photogrammetry and the development trend in the future. How to improve its positioning accuracy is the key to the research. Compared with the traditional air triangulation method, POS aided aerial triangulation can reduce or even eliminate ground control to a certain extent, which not only reduces the production cost, shortens the mapping period, but also improves the production efficiency. However, while studying the stability and reliability of POS system, it is very important to ensure the positioning accuracy of POS aided aerial triangulation. At present, the focus of this problem is mainly on the influence of system error, and the purpose of error compensation is usually achieved by changing the adjustment function model. However, POS assisted aerial triangulation is a joint adjustment between non-photogrammetric observations and photogrammetric observations, and the observed values involved in the adjustment are no longer a single image point coordinate. Also includes GPS (Global Positioning system) measured line elements and IMU (Inertial Measurement Unit) measured angular elements. Therefore, how to reasonably determine the weights of various types of observations will directly affect the quality of the adjustment results. Therefore, in the face of the diversity of observation value types, it is necessary to correctly and reasonably determine the random model of POS auxiliary space three. In this paper, the traditional weight determination method and the posterior variance component estimation method are used for the POS system observation values and the photogrammetric image point coordinate observations after the system error correction, respectively. Based on. Net platform, the program of POS data aided beam normal space triple-adjustment is compiled, and the experimental research is carried out in a certain area. The influence of different weight setting methods on the accuracy of adjustment results is analyzed. The experimental results show that there is little difference in the accuracy of the encryption point between the traditional weight determination method and the conventional measurement precision method, and the posterior variance component estimation method is used to adjust the error. The precision of the solution is much higher than that of the traditional weight determination method. However, using the same posterior variance component estimation method, but taking different initial weights will affect the convergence rate of the iterative calculation; when three different posteriori estimation methods are used, the convergence rate of the iterative calculation will be affected by using the same posteriori variance component estimation method. The precision of encryption point is slightly different, but because the Helmert estimation is always unbiased, and the process of Fostner estimation and Ebner estimation is from biased to unbiased, the number of iterations is different. The Helmert estimate and Ebner estimate iterate 8 times 11 times and 15 times, respectively. No matter which weight determination method is used in the adjustment process, the Z direction accuracy of the obtained encryption point is always lower than that of the XY direction. The necessity and validity of posterior variance component estimation in POS assisted beam method are verified by experiments.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：P231.5

【参考文献】