【作者】 唐文静；

【导师】 郝燕玲；

【作者基本信息】 哈尔滨工程大学 ， 导航、制导与控制， 2009， 博士

【摘要】 地理空间数据的获取在GIS工程里占有很重要的地位,同一地区的空间数据往往被不同部门重复采集,这不仅造成了人力、财力的巨大浪费,还引发了空间数据的二义性等问题,给GIS部门之间的数据共享和数据集成带来极大困难。解决这一问题最有效的方法是多源地理空间数据融合技术。二十一世纪是海洋的世纪,为满足我国海洋开发及沿海城市信息化建设的需求,本文围绕海陆地理空间矢量数据的融合技术,分别针对不同来源的海图、陆图在坐标、投影方式、几何数据、要素编码等方面存在的差异及其融合结果的不确定性等问题进行了研究。主要的研究内容包括：总结多源地理空间数据融合技术的研究内容及处理过程,讨论在实施多源地理空间数据融合前要解决的问题,并就海陆地理空间矢量数据融合的坐标系统一、投影方式统一等问题给出具体的实现过程。这些是研究海陆地理空间矢量数据融合的前提。同一地物在不同来源的地图上通常存在着差异,其识别或匹配对于多源地理空间数据融合来说很关键。借鉴空间相似性理论,基于人眼综合已有信息来识别同名实体的思想,本文提出基于空间相似性的实体匹配算法。该算法将实体作为一个整体看待,采用加权平均法来综合实体的位置、形状等特征的相似度,各指标权重依据视觉原理和人眼识别图形的特点来确定,进而根据获得的总相似度大小确定匹配实体。其中面实体的匹配,其指标计算引入了计算机视觉和模式识别中的方法。实验结果表明该算法能得到与人眼判断一致的结果,具有良好的稳定性和可靠性,且与其它算法比较,精度与召回率有明显提高。这是解决几何数据融合的基础。在同名实体匹配的基础上,为解决同名地物表达不一致的问题,综合不同来源数据的点位精度差异的影响,提出一种基于多评价因素的要素合并变换算法。分析确定影响合并变换的三大主要评价因素,并将其综合来确定要素的可信度,进而对要素位置进行加权平均来获得合并变换后的位置。结合海陆图的部分要素对该算法进行检验的结果表明,提高了要素的空间位置合并变换质量。这是解决几何数据融合的关键。研究多源地理空间数据融合中要素的编码融合问题。在阐述要素分类编码的原则和方法的基础上,提出融合的要素分类编码的原则和步骤,在此原则指导下,分析海陆图要素编码的差异,并解决编码融合过程中的要素层转换、同名要素统一编码等关键问题,实现海陆图要素编码的融合。数据融合的目的是为了提高融合后的信息量,信息不确定度的降低就相当于信息量的增加。因此在论文的最后,就多源地理空间矢量数据融合结果的不确定性进行分析。剖析矢量数据不确定性产生的各种原因,及其不确定度传播定律,在此基础上建立单源数据的不确定性模型,并通过多源矢量空间数据不确定性的联合模型建立数据源的不确定性与最终融合结果中不确定性的相互关系,以此来评定多源地理空间矢量数据融合质量。更多还原

【Abstract】 The gathering of geospatial data is very important in GIS applications. The same spatial data is sometimes colledted by different departments, which will cause the waste of human and financial resources, and brings data ambiguity. These problems bring many difficulties to data share and data integration between different GIS departments. An effective way to solve this problem is the geospatial data conflation technique. The 21th century is the ocean’s century. To satisfy the need of coastal economy development and coastal city information, conflation of geospatial vector data from a sea chart and a topographic map is studied in this paper. It solves the differences in coordinate systems, projections, geometries, codings etc and uncertainty of conflating results.General process of geospatial data conflation from multi-sources is summaried. Then the problems to be solved before putting geospatial data conflation into practice are discussed. And the process of uniting coordinate systems and projections of a sea chart and a topographic map is presented. These are the preconduction of the research on geospatial data conflation from multi-sources.Disparities of features that represent the same real world entity from different sources usually occur, thus their identification or matching is crutial to map conflation. Based on the spatial similarity theory and motivated by the idea of identifying the same entity through integrating known information by eyes, an entity matching algorithm is proposed in this paper. Regarding the entity as a whole, the total similarity is obtained by integrating positional similarity, shape similarity etc with a weighted average algorithm. And the weights are obtained based on vision theory and the characters of identifying graphics by eyes. Then the matching entities are obtained according to the maximum total similarity. Test results are consistent with human intuition, which show the stability and reliability of the proposed algorithm. Compared with other algorithms, precision and recall of the proposed algorithm are obviously improved. This is the base of solving the geometry conflation.Based on the matching of same entity, in order to solve their conflict and to synthesize the influence of elements precision on different source maps, an element adjusting algorithm based on multi-evaluation factors is proposed. Three primary evaluation factors are analyzed, and the element reliability is gained by integrating the three factors. Then the adjusted position is obtained with a weighted average algorithm. Some same elements from a sea chart and a topographic map are utilized to test the proposed algorithm. The result shows that the quality of areal element adjusting is improved. This is crucial in solving the geometry conflation.Coding conflation is also studied in this paper. Based on the explanation of principles and methods of element classification and coding, the principles and steps of conflated element classification and coding are presented. Guided by these principles, the differences of element coding from a sea chart and a topographic map are analyzed. Some key problems such as element layer transformation and the same element coding uniting are solved, and the coflation of coding is realized.The aim of geospatial data conflation is to improve the information content after conflation. The decrease of information uncertainty means the incerease of information content. Therefore, in the final chapter, the uncertainty of the conflation results is analyzed. The reasons that causing vector data uncertainty and the law of spreading uncertainty are introduced. Then the model of data uncertainty from single source is founded. And the relation of uncertainty from single source and from conflated results is constructed accoding to the unite model of uncertainty from multi-sources. This can be used to evaluate the quality of geospatial data conflation.更多还原