【作者】 孙笑古；

【导师】 李满春； 刘永学；

【作者基本信息】 南京大学 ， 地图学与地理信息系统， 2011， 博士

【摘要】 国家土地督察制度的建立为保护18亿亩耕地红线、规范地方政府合法利用土地资源提供了有力保障。然而,由于土地督察机构人员少、督察区域广,采取“人眼看、实地查”的传统手段难以担负全覆盖的督察重任,必须借助遥感信息技术,才能提高土地督察工作效能和威慑力量。目前,不断完善的多源异构的空间数据集成技术为进行土地督察行业数据快速建模,实现土地督察地理信息服务平台提供了技术支撑；同时,高速发展的遥感影像自动分析、信息提取技术为快速准确提取土地督察区域中疑似违法用地提供了一种有效的技术手段。但是,一方面土地督察机构成立时间尚短,针对土地督察的行业数据建模研究很少,土地督察地理信息服务平台的建设处于起步阶段；另一方面土地督察中利用遥感影像进行违法用地监测时完全依赖目视解译,在目标地块提取、变化图斑生成等环节中工作量巨大。基于以上背景,本文在国家土地督察机构“在线土地督察系统应用研究”课题的支持下,进行了“基于地理信息服务平台的土地督察违法用地监测系统研究”。主要研究内容和结论包括：(1)对土地督察专题信息进行梳理,识别专题数据之间的兼容对象并建立关系,基于国土资源数据标准建立土地督察数据模型,使得土地督察专业建模可扩展、可追溯、可互操作。在此基础上,设计、建立了土地督察空间数据库,实现了土地督察空间数据库与国土资源部数据信息中心的无缝集成,同时也拟定了土地督察空间数据标准。(2)提出了基于SOA的土地督察数据发现、获取与可视化的实现机制,以土地督察空间数据库为基础,建立了土地督察地理信息服务平台。该平台集成了多源异构数据,生产信息服务数据,能够为土地督察的各种应用系统提供标准化的空间数据服务。(3)在土地督察地理信息服务平台基础上,论文针对土地督察中使用的高分辨率影像的特点,结合违法用地中常见的线状、面状地物的主要特征,发展了多套高分辨率遥感影像中图像分割、信息提取算法,显著提高了土地卫片执法检查中变化图斑提取的效率。本文从土地督察实际工作出发,在已有国土资源信息化工作的基础上,实验性地构建了土地督察空间数据库,提出土地督察空间数据标准,基于SOA研究构建了土地督察地理信息服务平台。改进遥感影像分割和目标地物提取算法,进而构建土地督察违法用地监测模型,并以NET、Flex和ArcGIS Server作为开发平台和GIS服务平台,设计实现了土地督察违法用地监测原型系统。研究为探索土地督察信息化建设做出了贡献,同时为土地卫片执法检查中变化图斑的自动提取和违法用地类型的辅助判别方法研究提供了借鉴。更多还原

【Abstract】 In the context of the outstanding land issues and the increasingly serious situation of illegal land use, land supervision system provides a strong guarantee to regulate local government on the legal use of land resources. However, the traditional means, "people watch, field investigation" can hardly fulfill the task of land supervision. New information technology must be employed to achieve a fundamental change in the way of land supervision, improving the effectiveness and deterrent. Meanwhile, the multi-source heterogeneous spatial data integration technology can realize the data modeling for land supervision related information and a geographical information service platform. Technologies including remote sensing images automatically recognition, extraction of the changed features provide an effective technical means for a quick, accurate, large-scale extraction of an area of land inspectors suspected illegal use of land to.Of interest in this research is the illegal land-use monitoring system in land supervision based on geographical information service platform involving an important operation--"legal inspections of land use images". Legal inspections of land use images is a job of hard burden, low productivity, poor time efficiency due to a large amount of work including high-resolution remote sensing data collection, data acquisition, data management, data analysis and data processing, limiting the implementation of land supervision work. This research analyzes the essential problems met in the practice such as wide range of land supervision, a large volume of image data and a lot of intervention by human. Systematic solutions are proposed, with the focus on the construction of massive spatial data unified exchange platform, automatically recognition and extraction of illegal land use parcels. Following are major research questions and results:1) All kinds of information in the study area, especially the land supervision thematic information, are well organized. Compatibility among thematic data is identified and corresponding relationships are built. Land supervision data model is constructed based on the land industry data model, making the industry modeling extendable, traceable and friendly with interoperation. A spatial database satisfying land supervision is built, realizing the seamless integration of land supervision spatial database and information center of State Land Resource department. A land supervision spatial data standard is developed as well.2) A system scheme including online land supervision data discovery, data acquisition and visualization based on SOA is proposed. A geographical information service platform for land supervision is built based on the land supervision spatial database, which integrates multi-source heterogeneous data and producing information service data, providing various standard spatial data service for different applications associated with land supervision.3) Based on the geographical information service platform, methods to automatically extract the changed parcels is proposed based on the high resolution images used in land supervision and feature characteristics of common illegal land use. A decision-making support system assisting land supervision is constructed based on the modeling of discrimination procedure of common illegal land use types, to assist online processing and analyzing high-resolution remote sensing images related to "illegal land use inspection" and "legal inspections of land use images"This research, based on the current work of land informationization and practice of land supervision business, constructs a multi-level basic geographic database for land supervision through an empirical study, develops a land supervision spatial data standard and constructs a unified information platform for land supervision based on SOA. Algorithms related to recognition and extraction of changed parcels in multi-time series remote sensing images are improved, based on which a illegal land-use monitoring model assisting land supervision is built. A prototype of illegal land-use monitoring system for land supervision is designed and implemented relying on.NET, Flex and ArcGIS Server, as well as other GIS service platforms. This research contributes to the informationization of land supervision and methods associated with extraction of changed parcels through remote sensing images.更多还原