【作者】 曹闻；

【导师】 朱述龙；

【作者基本信息】 解放军信息工程大学 ， 摄影测量与遥感， 2011， 博士

【摘要】 传统的GIS处理的是静态空间数据,只能保留现实世界的一个瞬态,当数据发生变化时,一般用新数据替换旧数据,形成另一个瞬态,旧数据则不复存在,因而无法对空间对象的动态变化进行处理。为了准确地跟踪空间数据的动态变化,同时满足现实世界的不同应用需求,迫切需要设计恰当的、面向应用的时空数据模型,实现空间数据与时间信息的有机组织、高效管理及灵活使用。近年来,随着时空数据的广泛应用,相应的时空数据模型也相继而出,并逐步成为了当前具有重要理论和应用价值的研究热点。论文在现有研究成果基础上,从面向应用的角度提出了基于马尔可夫链的时空数据模型,通过对浮动车时空数据和地理时空数据的应用验证了该模型的有效性、实用性和通用性。论文的主要研究内容及创新点可以概括为:⑴系统地研究了现有时空数据模型的基本原理,通过空间语义、时间语义和时空语义等方面的对比分析,梳理出各个时空数据模型的优缺点,为时空数据模型的进一步研究提供了基础借鉴。⑵深入研究了地理对象时空变化的内部运行机理及其外在的空间变化特性,在现有时空数据模型的基础上,针对地理对象时空变化的无后效性、短时平稳性和误差特性等三种特性,从面向应用的角度提出了一种基于马尔可夫链的时空数据模型。该数据模型采用面向对象的技术,引入状态转移和时空粒度思想,有效地集数据模型和数据压缩为一体,集成了序列快照模型、基态修正模型和时空立方体模型等时空数据模型,从而提高了时空数据模型的可用性和通用性;同时基于隐马尔可夫模型构建了面向应用的统计分析模型,有效地描述了地理对象的时空演变,为时空数据面向应用建模提供了新的技术手段。⑶结合智能交通系统中浮动车时空数据的特点,利用基于马尔可夫链的时空数据模型思想构建了运动对象时空数据模型,并设计了城市实时交通信息发布原型系统,以此验证了基于马尔可夫链的时空数据模型的有效性和可用性。同时,针对原型系统中的数据采集、数据处理以及信息提取等重要技术环节分别提出了面向实际道路网络的浮动车自适应采样算法、基于短时预测的在线地图匹配算法、基于Hausdorff距离相似性测度的离线地图匹配算法和面向动态导航的浮动车交通拥挤判别算法,通过仿真试验验证了新算法的有效性和优越性。⑷根据遥感影像、数字矢量地图、DEM等地理时空数据多源、多分辨率、多时相、异构的特点,利用基于马尔可夫链的时空数据模型实现了地理时空数据的无缝衔接、高效组织、统一管理、快速查询和综合应用,为地理时空数据的一体化应用提供了技术支撑,同时验证了基于马尔可夫链的时空数据模型的可用性、有效性和通用性。更多还原

【Abstract】 Traditional GIS can only handle static spatial data and keep a transient state of the real world. Generally, historical data is replaced with fresh one to form another transient state, and we can not deal with dynamic change of spatial objects. An application oriented spatio-temporal data model is strongly needed to realize organic organization, efficient management and flexible usage of spatio-temporal data for tracking dynamic change of spatial data and meeting different application requirement of realistic world.Nowadays, spatio-temporal data has already extensively applied in different fields and corresponding spatio-temporal data model has also successively been put forward, which gradually becomes a hot research point with important theoretical value and application worth. This paper proposed a novel Markov-based spatio-temporal data model in terms of application based on the existing research results. We have verified the validity, practicability and universality of the model with the application in floating car spatio-temporal data and geographic spatio-temporal data. The main contribution of the dissertation is listed as follows:⑴The basic principle of existing spatio-temporal data model was systematically studied in the thesis. The merits and shortcomings of each spatio-temporal data model were listed out, and the universal problems of existing spatio-temporal data models were pointed out by analyzing spatial semantics, temporal semantics and spatio-temporal semantics. All these work provided a basic experience for the further spatio-temporal data model research.⑵The geography spatio-temporal objects and their space movement characteristic were analyzed. On the foundation of existing spatio-temporal data models, a noval spatio-temporal data model based on Markov chain was proposed. The data model absorbed the state transfer thought and spatio-temporal grain based on the object-oriented technique. Data model and data compression were efficiently gathered for the integral whole, and the sequence snap shot model, the ground state correction models and spatio-temporal cube model were integrated together. Thus the usability and applicability of spatio-temporal data model was significantly raised. The statistics analysis model was established for real application, which described the spatio-temporal evolvement of geography object effectively and provided a new technique means for spatio-temporal data modeling.⑶Combined with the characteristics of floating car spatio-temporal data in intelligence transportation system, a spatio-temporal data model of dynamic object was established utilizing spatio-temporal data model based on Markov chain thought. And a urban real time traffic information releases prototype system was designed to verify the usability and usefulness of the spatio-temporal data model based on Markov chain. Meanwhile, a series of new algorithms were put forward, which include the floating car adaptive sampling method oriented to actual road network, the on-line map matching algorithm based on the short-term prediction, the off-line map matching algorithm based on the similarity measure of Hausdorff distance and the floating car traffic congested status distinguishing method oriented towards dynamic navigation. The simulation and experimental results demonstrated the efficiency and superiority of these novel algorithms aimed in data collecting, data processing and information extraction of prototype system.⑷According to the multisource, multiresolution, multitemporal and isomerous properties of the geographic spatio-temporal data as remote sensing image, vector digital map and DEM, the spatio-temporal data model based on Markov chain was taken to realize the seamless link up, efficient organization, unify management, rapid search and synthesis application for the graphic spatio-temporal data. This work further provided a technical support for the geographic spatio-temporal data application and verified the usability, usefulness and adaptability of the spatio-temporal data model based on Markov chain at the same time.更多还原