项目名称: 基于几何代数的时空场数据特征解析与结构匹配方法研究
项目编号: No.41201377
项目类型: 青年科学基金项目
立项/批准年度: 2013
项目学科: 地理学
项目作者: 俞肇元
作者单位: 南京师范大学
项目金额: 25万元
中文摘要: 地理时空数据具有大尺度、多变量、边界不规则等特点。支持具有不规则边界的多维地理时空场数据的一体化表达、建模与模拟是地学分析与GIS研究的热点。本研究紧密跟踪数学、物理学等基础研究的前沿领域,引入张量与几何代数,构建基于多重向量的多维时空场的统一组织与表达方法,实现对标量场、矢量场、格网场以及任意混合场的压缩存储、快速检索与透视、切片等操作。基于几何代数算子,建立多维时空场基本表征参数集,并分别构建数据驱动及特征模板驱动的多维不规则时空场特征解析方法。研究多维时空场数据的Clifford卷积与相似性度量的快速算法,建立结构匹配模板及核函数的自适应生成方法,进而实现对多维时空场中显式或潜在的结构特征的匹配与重构。本项目有助于突破现有多维时空数据特征解析与结构匹配方法的局限性,为具有普适性的地学多维时空数据分析提供理论框架与实现途径,并可为地理现象发展演化过程的表达、建模与模拟提供技术支撑。
中文关键词: 时空场数据;几何代数;特征提取;结构匹配;时空分析
英文摘要: With the development of the global earth observation, there are already accumulated large amounts of geographic spatio-temporal fields data. These data are usually spans large spatial and temporal scales and have lots of different variables. New data types such as multidimensional pointclouds, irregular grid fields, etc, which usually have different dimensions and irregular boundary should be processed efficiently. It has already been a research hotspot of GIS and disciplines that developing new processing methods to support multidimensional geographic space and time data integration, expression, modeling and simulation with complex multidimensional spatio-temporal data. This study learns from the research frontier area of fundamental basic researches such as mathematics, physics, etc., to find the solution to the feature extraction and structure matching problem of complex spatio-temporal fields data. Tensor algebra and geometric algebra are introduced in the research. The unified organization method, the compressed storage mechanism, rapid retrieval and perspective, sliced operation of complex spatio-temporal data will first be designed. The proposed method can integrate multidimensional spatio-temporal fields, including scalar fileds, vector fields, grid fields and their mixtures, in a unified multivector st
英文关键词: Spatio-temoral field data;Geometric algebra;Feature extraction;Structure Matching;Spatio-temporal analysis