基于智能对象的决策支持系统体系结构研究

分析了决策问题及其求解过程 ,揭示了传统IDSS体系结构的不足 ,阐述了应用面向对象的结构化知识表达构造智能对象 ,提供了智能决策过程支持的方法和新的IDSS体系结构 ,并在农业空间决策信息系统的实践中进行了验证     

集成高度和彩色纹理特征的影像目标模糊聚类识别方法

提出一种基于高度和彩色纹理信息的目标识别方法 ,其目的是提取具有相对高度的地物。由于传统二维纹理分析的方法往往不能有效地提取这类目标 ,本文把高度、色彩和纹理信息有机地结合起来 ,提高了识别率。依据影像提取结果 ,探讨了树林对森林覆盖地区自动空中三角测量选点的负面影响 ,同时建议采用数据分组的方案来改善森林覆盖地区自动空中三角测量的精度 ,并用对比实验证明了其有效性。     

Seasonal Variation of the East Asian Subtropical Westerly Jet and Its Association with the Heating Field over East Asia

The structure and seasonal variation of the East Asian Subtropical Westerly Jet （EAWJ） and associations with heating fields over East Asia are examined by using NCEP/NCAR reanalysis data. Obvious differences exist in the westerly jet intensity and location in different regions and seasons due to the ocean-land distribution and seasonal thermal contrast, as well as the dynamic and thermodynamic impacts of the Tibetan Plateau. In winter, the EAWJ center is situated over the western Pacific Ocean and the intensity is reduced gradually from east to west over the East Asian region. In summer, the EAWJ center is located over the north of the Tibetan Plateau and the jet intensity is reduced evidently compared with that in winter. The EAWJ seasonal evolution is characterized by the obvious longitudinal inconsistency of the northward migration and in-phase southward retreat of the EAWJ axis. A good correspondence between the seasonal variations of EAWJ and the meridional differences of air temperature （MDT） in the mid-upper troposphere demonstrates that the MDT is the basic reason for the seasonal variation of EAWJ. Correlation analyses indicate that the Kuroshio Current region to the south of Japan and the Tibetan Plateau are the key areas for the variations of the EAWJ intensities in winter and in summer, respectively. The strong sensible and latent heating in the Kuroshio Current region is closely related to the intensification of EAWJ in winter. In summer, strong sensible heating in the Tibetan Plateau corresponds to the EAWJ strengthening and southward shift, while the weak sensible heating in the Tibetan Plateau is consistent with the EAWJ weakening and northward migration.     

基于ArcGIS Server的服务器对象扩展方法研究

针对ArcGIS Server现有功能无法满足一些特殊业务逻辑的问题,本文对Arc GIS Server服务器扩展方法进行了研究。阐述了ArcGIS Server面向服务的底层架构,在此架构下对三种主要GIS服务器扩展方法进行对比分析,得出服务器对象扩展是最可行的方法。最后对该方法的模型结构和实现流程进行研究,并通过具体的项目实例进行说明。     

高分一号遥感影像地质灾害信息提取方法研究

山区地质灾害发生频繁,受灾范围大,危害严重。提取地质灾害信息、估算受灾情况,对救灾工作极为重要。传统的遥感人工解译方法速度慢、效率低;计算机自动解译速度快但解译精度受影像质量影响大,大范围地区很难建立起普适性的解译模型。本文利用洮南市西部高分一号遥感数据,结合DEM生成三维影像,建立解译标志,解译基础地质信息,分析地质灾害成因;根据研究区泥石流地质灾害的主要影像特征,提取影像的分类属性,基于面向对象的分类方法,建立信息提取模型,快速提取出地质灾害敏感区域;再进行人机交互,精确提取出地质灾害的类型和范围,通过野外验证,该方法十分可靠,为大范围地质灾害信息快速提取和灾后救援提供科学依据。     

冬季东亚高空副热带急流和温带急流协同变化与中国南方地区降水的关系

利用NCEP/NCAR再分析数据和中国台站降水资料研究冬季东亚高空副热带急流和温带急流协同变化特征及其与中国南方地区降水的关系,发现冬季东亚高原急流与温带急流同期反向协同变化特征最为显著。即高原急流增强,同时温带急流减弱(SW型)和高原急流减弱,同时温带急流增强(WS型)。当高原急流增强(减弱)而温带急流减弱(增强)时,中国南方地区降水显著增加(减少)。合成分析表明,不同急流协同变化型态下冷暖空气活动特征存在较大差异,高原急流与温带急流的反向协同变化可以真实反映与冬季中国南方地区降水相关联的冷暖空气活动特征,进而导致不同降水形态的产生。     

鲁中南一次暴雨过程的特征和成因分析

利用常规气象观测资料、NCEP 1°×1°逐6h的再分析资料、GPS可降雨资料和地闪定位资料,对2013年7月4—5日山东中南部出现的暴雨到大暴雨天气过程的不同阶段特征和成因进行了分析,结果表明:本次暴雨过程是冷暖切变线所引发,暴雨的水汽源于南海;低空急流的强弱和水汽通量的大小呈正相关;暖切变线暴雨的雨强、影响范围和持续时间明显大于冷切变线暴雨;冷、暖切变线GPS可降雨量表现不同,前者短时间内增幅大,地面强降雨在峰值出现1h后发生,对地面降雨变化反映较敏感,后者强降雨出现前8h可降雨量快速上升,可降雨量峰值对应地面降雨大值,对地面降雨变化反映不敏感;冷切变线对流性更强,地闪频次为暖切变线降雨的2倍,正地闪频次为暖切变线降雨的1/2,负地闪频次为暖切变线降雨的2倍;降雨云团正、负地闪活跃程度呈反相位关系,正、负地闪的变化能很好地反映强降雨的变化;冷、暖切变线动力结构不同,前者物理量场从低层到高层向北倾斜,后者则为垂直分布;冷切变线上升运动区较暖切变线深厚。     

影像与激光雷达数据的城市地物信息提取

针对城市地物信息提取中地物边界难以确定、分类精度不高的问题,该文提出一套综合利用影像及激光雷达点云高程信息的面向对象分类方法。在分割中,各类地物的最佳分割尺度由监督法分割精度评价确定,最终分割结果利用粒度理论下的分割尺度综合方法进行合成,能兼顾不同地物最优分割尺度,获得准确地物边界;在分类中,采用ReliefF特征选择算法度量从影像及点云数据提取的对象特征重要度,选择最佳特征组合,并采用多分类器组合方法进行分类,以消除Hughes现象,提高分类精度。选择德国斯图加特市两块实验区进行分类实验,结果表明:该方法有利于提高大范围城市地物精细信息提取的精度和效率,具有较高的应用价值。     

支持向量机的车载雷达点云目标识别

为了进一步研究移动测量系统的数据处理问题,该文根据点云的基本特征,归纳了由7个特征构成的点云原始特征向量,在此基础上,结合语义环境构建了由17个特征构成的点云扩展特征向量,并采用支持向量机模型对车载LiDAR点云进行行道树点云识别的一系列实验。实验中采用粒子群优化算法和遗传算法对支持向量机进行参数寻优;采用不同特征向量和不同数目样本对点云进行学习和目标识别;分析了特征向量的学习曲线和识别精度。实验结果表明,支持向量机模型能够在行道树点云识别中取得较高的精度。     

Tessellations in GIS: Part II–making changes

Abstract

We attempt to describe the role of tessellated models of space within the discipline of Geographic Information Systems (GIS) – a speciality coming largely out of Geography and Land Surveying, where there was a strong need to represent information about the land’s surface within a computer system rather than on the original paper maps. We look at some of the basic operations in GIS, including dynamic and kinetic applications. We examine issues of topology and data structures, and produced a tessellation model that may be widely applied both to traditional “object” and “field” data types. The Part I of this study examined object and field spatial models, the Voronoi extension of objects, and the graphs that express the resulting adjacencies. The required data structures were also briefly described, along with 2D and 3D structures and hierarchical indexing. The importance of graph duality was emphasized. Here, this second paper builds on the structures described in the first, and examines how these may be modified: change may often be associated with either viewpoint or time. Incremental algorithms permit additional point insertion, and applications involving the addition of skeleton points, for map scanning, contour enrichment or watershed delineation and simulation. Dynamic algorithms permit skeleton smoothing, and higher order Voronoi diagram applications, including Sibson interpolation. Kinetic algorithms allow collision detection applications, free-Lagrange flow modeling, and pen movement simulation for map drawing. If desired these methods may be extended to 3D. Based on this framework, it can be argued that tessellation models are fundamental to our understanding and processing of geographical space, and provide a coherent framework for understanding the “space” in which we exist.