一种稳健性准则及相应的稳健估计

研究了已有的各种稳健性度量，根据稳健性的经典定义，本文建立了一种衡量稳健性的准则，根据这一准则，研究了在测量中常用的两种误差模型下的稳健结构。结果表明，在这一个准则和随机误差模型下的最优稳健估计，是李德仁教授提出的验后方差法；在这一准则与均值移动误差模型下的最优稳健估计，是具有均方误差最小的稳健估计。     

粗差验后方差的无偏估计与最优稳健估计

在正态粗差假设下导出了粗差验后方差的无偏估计,对误差工膨胀模型和误差均值移动模型,两者的无偏估计公式是相同的。这证明了李德仁验后方差的朱建军方差不是无偏的。由于偏方定义的彭方法是正态粗差假设下的最优稳健估计。     

污染误差模型下的测量数据处理理论

本文首先研究了污染误差模型的各种具体的误差表示形式,然后研究了误差服从污染误差模型时的平差准则。指出,当误差服从污染误差模型时选择均方差作为估计准则是合理的,并且符合传统的测量误差处理的观念。最后,推导了误差服从污染误差模型时,均方误差准则下的最佳估计,从而建立以均方误差准则为基础的污染误差模型下的测量数据处理理论。     

污染模型下的最优估计

在目前的抗差估计理论研究中,抗差估计的建立与模型误差没有关系,主要是依据抗差准则设计相应的权函数,但权函数的选择带有主观性,因而抗差估计是经验的。本文就合适于测量数据处理的污染误差模型,提出了未知参数向量和方差的最小均方误差的抗差估计,探讨了最小均方误差估计的计算问题。     

方差分量估计的抗差总体最小二乘算法

应用等价权函数建立总体最小二乘抗差估计算法,算法的有效性受到随机模型的影响。为减小随机模型误差对抗差算法的影响,提出方差分量估计确定总体最小二乘随机模型的思想;利用等价权函数对观测元素进行迭代定权,分类确定等价权函数的域值,提高权函数抗差算法的有效性。最后,通过算例验证算法可行性,结果表明本文算法是有效的。     

GIS中曲线误差的模型与试验研究

提出了描述曲线整体误差的随机过程模型．用过程的数字特征函数定义了曲线的局部误差指标．用过程的积分定义了曲线的局部和整体误差指标．阐述了各指标的概率意义与几何意义．设计了过程模型的数字化试验，并提取了过程的样本曲线。通过对样本曲线的统计分析．得到了各误差指标的估计值。     

Real-time estimation of satellite clock offset using adaptively robust Kalman filter with classified adaptive factors

In order to estimate the satellite clock offset in a real-time mode, a new algorithm of adaptively robust Kalman filter with classified adaptive factors for clock offset estimation is proposed. Compared with standard Kalman filter clock offset model, the new method can detect and control outliers and clock jumps automatically in real-time. Moreover, the clock model parameters, which contain the clock offset, clock speed and clock shift, are classified to decide the adaptive factors in the new model. Thus, clock jumps with different characteristics can be distinguished more effectively. Meanwhile, the dynamic noise characteristics of clock offset series are used for stochastic modeling. An actual numerical example is presented, which shows that the proposed filter can give a better performance than other commonly used filters.     

平差系统的模型误差及其识别方法研究

论述了模型误差影响参数估值的一些理论问题,指出了随机模型误差和函数模型误差之间的相互作用和转化。为讨论平差系统最优模型的选取,给出了与现有文献将模型误差纳入平差系统的思路不同的一个估计和识别模型误差的理论基础公式,由此导出了相应的实用公式,给出了平差系统模型的优选方法。     

Statistical analysis of signal measurement in time-of-flight cameras

Three-dimensional imaging systems have evolved significantly in the last two decades due to increase in demand for tasks in the field of close range photogrammetry. The fast and growing need of 3D imaging devices has given rise to range image technology, especially time-of-flight (TOF) cameras, that provide direct measurement of distance between the camera and the targeted surface. A significant advantage of TOF devices over traditional range data sensors is their capability to provide frame rate range data over a full image array. In phase shift TOF cameras, phase shift sampling of the received signal is used to measure amplitude, phase and the offset (intensity) of the received signal. As a result, the quality of the measurement of these sensors depends heavily on signal-to-noise (SNR) of the incoming signal and the subsequent processing algorithms. A detailed understanding of the statistical distributions of the measurement parameters is crucial for accurate distance measurement analysis especially in low SNR scenarios. In this paper, we provide explicit noise models for the three parameters of amplitude, phase and intensity. The proposed stochastic model helps in investigating the effect of noise on signal and classifying range data reliability in TOF cameras. The model is used for prediction of errors in a TOF camera under various SNR conditions. Experimental verification confirms the validity of the model using real data for range error classification under different noise conditions.     

基于EIV模型的稳健估计

EIV（error-in-variables）模型同时考虑观测向量和系数矩阵的误差,自提出以来便得到广泛应用。目前针对EIV模型的整体最小二乘解法（TLS）假设观测值仅含有偶然误差,当观测值存在粗差时其解并不是最优的。文中通过选定合适的权函数,结合加权整体最小二乘迭代算法,导出基于EIV模型的稳健整体最小二乘迭代解法（RTLS）。线性拟合实验表明,文中方法能对粗差进行定位,且估计量受粗差影响较小,具有稳健性。     

Error model of curves in GIS and digitization experiment

IntroductionCurves are the fundamental geometric elementsin GIS. The uncertainty of their positions has ani mportant influence on those outputs from GISandis the reliable basis for reasonably evaluatingthose outputs .In contrast to measurement error theor…     

Error model of curves in GIS and digitization experiment

A stochastic error process of curves is proposed as the error model to describe the errors of curves in GIS. In terms of the stochastic process, four characteristics concerning the local error of curves, namely, mean error function, standard error function, absolute error function, and the correlation function of errors, are put forward. The total error of a curve is expressed by a mean square integral of the stochastic error process. The probabilistic meanings and geometric meanings of the characteristics mentioned above are also discussed. A scan digitization experiment is designed to check the efficiency of the model. In the experiment, a piece of contour line is digitized for more than 100 times and lots of sample functions are derived from the experiment. Finally, all the error characteristics are estimated, on the basis of sample functions. The experiment results show that the systematic error in digitized map data is not negligible, and the errors of points on curves are chiefly dependent on the curvature and the concavity of the curves.     

顾及运动估计误差的“凝视”卫星视频运动场景超分辨率重建

在含有运动地物的“凝视”卫星视频影像超分辨率重建问题中,针对由于运动估计不准确导致的重建后运动物体存在“格网”和“拖尾”等问题,提出了顾及运动估计误差的运动场景超分辨率重建方法。首先,在MAP模型框架下分析了基于L1范数和L2范数的保真项对于运动估计误差的稳健性,引入稳健的M-估计作为保真项,自适应的减少运动估计误差对重建结果的影响;然后,分析了稳健估计条件下Tikhonov、TV、BTV正则项的重建效果;最后,提出了基于稳健估计的双边滤波超分辨率重建方法,用SkyBox和吉林一号的卫星视频数据验证本文方法。试验证明,本文方法对含有运动目标的卫星视频影像具有较好的重建效果,重建后动态和静态地物细节都得到提升,归一化方差和梯度能量指标都优于其他方法。试验表明了本文方法的可行性。     

基于随机过程的格网DEM精度场模型

针对目前DEM格网精度从理论上难于统一量化描述问题,提出一种基于随机过程的格网DEM（grid digital elevation model, grid DEM）精度场模型,该模型从随机过程理论入手,借助DEM的统一插值模型,建立了DEM两类误差（噪声误差和逼近误差）的数学表达,并分析了构成DEM误差的各个分量的数学意义。与已有的模型相比,该模型实现了传统DEM两类误差的统一描述,此外,该模型适用任意插值方式生成的格网DEM的精度评价。     

基于均值平移与模拟泛洪的图像分割算法

针对图像处理的特点，综合考虑执行速度与健壮性要求，提出了一种顺序搜索LUV空间候选点的均值平移实现方法。为了克服传统的基于特征空间分析的图像分割方法对像素点空间关系考虑不够充分的缺陷，借鉴Vincent ＆ Soille沉浸模拟分水岭算法思想，构造了一种与均值平移方法相适应的新的模拟泛洪方法执行图像分割。新方法比较符合人的主观感知，且易于容纳更多的特征以满足特定的图像分割需求。对实验用彩色图像的分割结果表明了该方法的有效性。     

基于均值漂移模型的粗差定值定位

针对粗差定值定位问题,该文从观测值与其估值的解析关系出发,提出了一种基于均值漂移模型的粗差定值定位研究思路和方法。首先利用二次投影公式和帽子矩阵的性质,推导出了参数分量和粗差估计值的另一种估计表达式,讨论了相应估计量的统计性质。然后,根据不同情况,构造了粗差检验统计量,并给出了相应的粗差检验步骤。最后,用一组包含不同粗差的模拟算例验证了该文方法的有效性。从理论上丰富了粗差探测的方法,扩大了均值漂移模型的应用范围。     

Stochastic modeling for time series InSAR: with emphasis on atmospheric effects

Despite the many applications of time series interferometric synthetic aperture radar (TS-InSAR) techniques in geophysical problems, error analysis and assessment have been largely overlooked. Tropospheric propagation error is still the dominant error source of InSAR observations. However, the spatiotemporal variation of atmospheric effects is seldom considered in the present standard TS-InSAR techniques, such as persistent scatterer interferometry and small baseline subset interferometry. The failure to consider the stochastic properties of atmospheric effects not only affects the accuracy of the estimators, but also makes it difficult to assess the uncertainty of the final geophysical results. To address this issue, this paper proposes a network-based variance–covariance estimation method to model the spatiotemporal variation of tropospheric signals, and to estimate the temporal variance–covariance matrix of TS-InSAR observations. The constructed stochastic model is then incorporated into the TS-InSAR estimators both for parameters (e.g., deformation velocity, topography residual) estimation and uncertainty assessment. It is an incremental and positive improvement to the traditional weighted least squares methods to solve the multitemporal InSAR time series. The performance of the proposed method is validated by using both simulated and real datasets.     

开窗分类因子抗差自适应序贯平差用于卫星钟差参数估计与预报

提出一种新的钟差算法--开窗分类因子抗差自适应序贯平差,即首先对一维钟差数据进行开窗处理,在窗口内利用抗差等价权削弱粗差影响,在窗口间构造自适应因子抵制钟跳异常,从而达到消除和削弱观测异常和状态异常的目的.针对不同星钟参数不符值描述不同的扰动特性,提出构造分类自适应因子来抵制钟差时间序列中的扰动异常.计算结果表明,新算...     

单位权方差的岭估计和压缩估计

本文提出了求单位权方差的有偏估计方法─岭估计和压缩估计，证明了在一定条件下，岭估计和压缩估计的均方误差要小于最小二乘估计的约方误差，提出了求岭参数和压缩系数的方法。     

均值漂移与卡尔曼滤波相结合的遥感影像道路中心线追踪算法

基于模板匹配的道路追踪方法是道路提取中较实用的一类方法,但传统模板匹配方法主要以相关系数作为相似性测度,对车辆、树荫等遮挡敏感,不适用于高分辨率遥感影像道路提取。针对这一问题,本文采用一种稳健的相似性测度,设计了一种基于均值漂移的道路中心点匹配算法,克服了传统模板匹配对遮挡敏感的缺点;然后运用卡尔曼滤波,实现高分辨率遥感影像道路中心线追踪。试验表明,该方法能够准确提取高分辨率遥感影像道路中心线,对车辆、树荫等遮挡具有稳健性。