Least-squares collocation with covariance-matching constraints

Most geostatistical methods for spatial random field (SRF) prediction using discrete data, including least-squares collocation (LSC) and the various forms of kriging, rely on the use of prior models describing the spatial correlation of the unknown field at hand over its domain. Based upon an optimal criterion of maximum local accuracy, LSC provides an unbiased field estimate that has the smallest mean squared prediction error, at every computation point, among any other linear prediction method that uses the same data. However, LSC field estimates do not reproduce the spatial variability which is implied by the adopted covariance (CV) functions of the corresponding unknown signals. This smoothing effect can be considered as a critical drawback in the sense that the spatio-statistical structure of the unknown SRF (e.g., the disturbing potential in the case of gravity field modeling) is not preserved during its optimal estimation process. If the objective for estimating a SRF from its observed functionals requires spatial variability to be represented in a pragmatic way then the results obtained through LSC may pose limitations for further inference and modeling in Earth-related physical processes, despite their local optimality in terms of minimum mean squared prediction error. The aim of this paper is to present an approach that enhances LSC-based field estimates by eliminating their inherent smoothing effect, while preserving most of their local prediction accuracy. Our methodology consists of correcting a posteriori the optimal result obtained from LSC in such a way that the new field estimate matches the spatial correlation structure implied by the signal CV function. Furthermore, an optimal criterion is imposed on the CV-matching field estimator that minimizes the loss in local prediction accuracy (in the mean squared sense) which occurs when we transform the LSC solution to fit the spatial correlation of the underlying SRF.     

探地雷达自适应天线的数值仿真

探地雷达系统中的自适应天线采用长条形偶极子单元天线构成等效蝶形天线,通过接通和断开相应单元天线,改变天线自身的等效张角,使其能达到改变天线在土壤中覆盖区域的目的。采用有限积分法对该天线在分层土壤环境下的覆盖区域以及单一土壤环境下的阻抗匹配能力进行了数值仿真,仿真结果表明,该天线对不同大小和形状的目标物体具有自适应能力。     

IHS方法在QuickBird数据融合中存在的问题及其改进

 针对IHS变换在QuickBird数据融合中存在的光谱扭曲问题，提出了利用Visual-Pan波段和线性加权匹配两种方法进行改进 ，并给出了Visual-Pan波段方法中系数α的最佳取值范围，以及线性加权匹配融合图像的空间特征与光谱特征达到最佳效果时Pan 权值(wPan)和I权值(wI)的最佳取值。结果表明，对于Visual-Pan方法而言，当0.2＜α＜0.25时，可以获得非常好的融合效果; 而 对于线形加权匹配方法而言，当wpan=3/4、wI=1/4时，融合图像的空间特征与光谱特征可以达到最佳效果。     

基于角点特征的近景影像匹配算法

利用了改进的Harris特征点提取算子从原始立体像对提取一系列特征点,采用了自适应三角形约束下的影像匹配方法进行近景影像匹配,试验结果表明其具有良好的可靠性。     

基于小波分量特征值匹配的高光谱影像分类

提出了一种基于小波分量特征值的高光谱影像分类算法。针对每个像素构建一个能反映该分量特征的函数，得到其特征值。再利用这些特征值与参考光谱的特征值进行匹配，从而对整幅影像实现分类。实验证明，该方法比传统的光谱角制图法和交叉相关系数法的分类精度有较大提高。     

U.K.USAGE OF CHEMOMETRICS AND ARTIFICIAL INTELLIGENCE IN QSAR ANALYSIS

A survey of members of the U.K.QSAR Discussion Group has been made to determine the extent ofuse and development of chemometric and artificial intelligence(AI)methods in the analysis ofmultivariate quantitative structure-activity relationship(QSAR)data in the U.K.Chemometric methodswere found to be well established in both industrial and educational establishments and there wassignificant method development occurring.AI methods were not employed to any great extent and thegeneral level of interest in these techniques was low compared to chemometric methods.A requirementfor more education in multivariate statistical methods and regression methods was indicated.A need fora user-friendly,comprehensive,commercially available multivariate statistical package containingmultivariate stability testing and regression diagnostic methods was identified.     

威特DI2002测距仪性能及其与ME－3000反射镜匹配研究

本文通过对威特ＤＩ２００２测距仪性能的测试。发现影响测距精度的主要因素是ＧＲＴ１０－ＧＰＨ１Ａ的反射镜系统。威特ＤＩ２００２测距仪匹配ＭＥ－３０００反射镜后，需加两项改正，但观测精度显著提高，实测表明达到与ＭＥ－３０００测距仪相同的精度。     

A CLASS-MODELLING TECHNIQUE BASED ON POTENTIAL FUNCTIONS

A probabilistic and distribution-free class-modelling technique is developed from potential functiondiscriminant analysis.In the multidimensional space of variables the class boundary is built either by thesample percentile of the probability density estimated by means of potential functions,or by the estimateof the ‘equivalent’ determinant of the variance-covariance matrix.The equivalent determinant is that ofa hypothetical multivariate normal distribution whose mean probability density was obtained by potentialfunctions.The bases of this modelling rule are evaluated by means of Monte Carlo experiments.Theresults on four datasets are used to measure the performances of this method,which equal and sometimesexceed the performances of parametric class-modelling methods based on linear and quadraticdiscriminant analysis which were used for comparison.     

利用遥感影像进行变化检测

首先分析了利用影像进行变化检测的紧迫要求和存在的困难 ,接着对变化检测的方法进行分类 ,提出将影像配准与变化检测整体同步解求的思想和在此基础上进行三维变化检测的可能性 ,并在文中概要介绍和比较了各种变化检测的 7种方法 ,最后展望了下一步应继续研究的方向。