分割图像上的区域及其空间关系的表示

在影像分析过程中，分割图像上的区域及其相互关系的表示是十分重要的。它是后继分析处理过程的基础。本文讨论了面向区域的游程编码表示方案，在此基础上给出了判断区域之间空间关系的算法，并对关系图在计算机中的存储做了研究。这些方法对计算机视觉和图像信息系统中的空间推理都是十分有用的。     

图像匹配的物体空间关系推理表达

物体空间关系指的是物体在欧氏空间中的邻近关系,根据图像中包含物体的邻近关系解决图像匹配的问题。本文首先基于对比机制训练物体块特征提取网络,构建物体块深度特征,该特征可以有效匹配不同图像中的相同物体块;其次,基于已有的先验图像数据推理表达图像中物体的空间邻近关系,构建场景物体空间邻近图;进而基于该空间邻近图计算场景图像对的空间邻近度,完成图像空间关系匹配。试验表明不匹配图像间的空间邻近度一般为0,而匹配图像间的空间邻近度一般大于0,本文空间关系匹配涉及多个物体间的相互关系,具有更强的稳健性,其匹配效果明显优于对比试验中的其他方法,可以高效稳定地完成图像匹配任务。     

Spatial relationship-assisted classification from high-resolution remote sensing imagery

Spatial information remains to be an important topic in geographic information system and in remote sensing fields, and spatial relationships have been increasingly incorporated into the image classification processes. Previous studies have employed multiple occurrences of spatial features (shape, texture, etc.,) to improve classification results. However, less attention has been focused on using higher-level spatial relationships for image classification. In this study, two novel spatial relationships, namely, maximum spatial adjacency (MSA) and directional spatial adjacency (DSA), were proposed to assist in image classification. The proposed methods were implemented to extract buildings, beach, and emergent vegetation land-cover classes according to their spatial relationships with their corresponding reference classes. The promising results obtained from this study suggest that the proposed MSA and DSA spatial relationships can be valuable information in defining rule sets for a more reasonable and accurate classification.     

基于植被分布特征的森林景观建模方法

在虚拟地理环境中如何快速地生成具有特定空间格局的森林景观是一个具有挑战的问题。现有的研究主要以模拟或随机的方式确定树木的位置,但难以反映每个树种的分布特征。本文提出了一种基于图的中性景观模型,根据指定的景观指数生成特定组成结构和空间配置的森林景观。模型将格网随机划分为若干不规则的区域,并将其转化为对应的区域邻接图,通过源节点的受限生长生成特定大小和形状的图斑。相比其他中性景观模型,该模型在可控性和灵活性上具有优势,能对每种类型的景观特征分别进行表达。最后根据模型生成的森林景观确定树木的位置与种类。本方法不仅能为植被场景的可视化提供合理的植被分布,而且能反映森林景观的空间分布特征,为虚拟森林环境中的林业规划决策以及植被分布规律认知提供一定的支持。     

一种顾及空间关系的遥感影像阴影检测算法

以城市区域内高大建筑阴影为研究对象,针对现有的阴影检测算法在复杂地物环境下检测精度和可靠性不高的问题,提出了一种结合颜色空间特征和空间关系的遥感影像阴影检测方法。首先,采用SLIC超像素算法对影像进行分割;然后基于Lab和HSI颜色空间构建初步检测条件,将阴影划分为阴影主体区域和待检测区域;最后,借助Canny边缘检测信息合并待判别区域内的超像素块,并利用阴影区域与造成干扰区域间的空间位置关系构建的检测条件进行判别。实验结果表明,该方法可以有效提高复杂地物环境下遥感影像阴影的检测精度和算法可靠性。     

A Method for Generating Contour Tree Based on Voronoi Interior Adjacency

IntroductionTopographic contour lines play an i mportantrole in the process of humans analyzing land-form. Particularly ,the spatial relations of con-tours are usually analyzed andinterpreted by hu-mans to recognize topographic features .In orderto model and represent the relationships betweencontour lines ,in 1963 Boyell and Reston pro-posed a contour tree generated by mapping con-tour lines as edges and interstitial spaces asnodes[1 ,2].There has been a great deal of previous workrepresenti…     

基于Voronoi内邻近的等高线树生成法

针对等高线树现有几何计算生成法和区域扩张生成法在判断包含关系方面存在的不足,提出了利用Voronoi内邻集来判断父等高线与子等高线之间“一对多”的直接包含关系,发展了一种基于Voronoi内邻近的等高线树生成方法。     

A method for generating contour tree based on voronoi interior adjacency

A contour tree is a good graphical tool for representing the spatial relations of contour lines and has many applications in map generalization, map annotation, terrain analysis, etc. A new method for generating contour trees by introducing a Voronoi-based interior adjacency concept is proposed in this paper. The immediate interior adjacency set is employed to identify all of the children contours of each contour without contour elevations. It has advantages over existing methods such as the geometric method and the region growing-based method.     

Similarity Measurements on Multi‐Scale Qualitative Locations

Qualitative representation of spatial locations and their similarity measurements are essential for the analysis of linguistic term‐based data. Existing methods have focused on the similarities of spatial relations and spatial scenes but have not considered the variations in geometrical representations and relations over scales. This study developed some new measures to assess the similarities of both single‐ and multi‐scale qualitative locations. Region‐ and cell‐based models were used to formalize qualitative locations of spatial objects with respect to multi‐scale frames of reference. The similarities were assessed by integrating the similarities of frames and qualitative relations. The frame similarity measures how two objects are compared considering the common elements that they occupy in the reference frames. Moreover, the similarity of qualitative relation measures how two relations relate two objects to the corresponding elements in the frames. The location similarities at a single level integrate the similarities of the frames and qualitative relations, whereas the location similarities at multiple scales incorporate the variations in qualitative locations over scales. These methods were used to assess location similarities concerning residential areas, roads, and lakes. The results indicated that the location‐based measurements can disclose the distributions of the similarities and that the cell‐based model is more accurate than the region‐based model.     

凹边形地标参照的主方向关系推理方法

空间推理是地球空间信息、人工智能、自然语言处理等相关领域的热点研究内容。主方向关系的表达与推理是空间推理的重要组成部分。针对空间方向关系矩阵对凹边形对象方向关系表达、推理的不足,研究依据目标对象与凹边形地标方向关系表达的不确定性,提出了凹边形地标参照的外部性主方向关系与内部性主方向关系推理方法。利用凹边形三等分仿射变换矩阵、自适应分割算法实现凹边形地标外部性、内部性参照对象表达;基于向量三角形法则与方向关系谓词,以方向关系的逻辑和、逻辑差合成代数运算推理2个地标、1个地标参照下目标对象与参照对象之间的主方向关系。通过实验与算例验证,证明凹边形地标参照下目标对象与参照对象的主方向关系推理可行性。     

利用小波变换的高分辨率多光谱遥感图像多尺度分水岭分割

为了减少仅用分水岭变换而导致的过分割问题,本文提出利用小波变换的多尺度处理方式用于融合后多光谱QuickBird图像的分割。整个分割过程包括多尺度图像表示、图像分割、区域合并和结果映射等过程。首先,依据原始图像的大小确定分解尺度并用小波变换产生各波段的低尺度图像。采用相位一致模型提取各近似系数的梯度,并逐尺度地融合各梯度图。分析不同尺度下的不同地物的局部梯度方差,以选择最佳的小波分解尺度。然后,通过移动阈值与扩展最小变换,利用多层次标记提取方法标记均质区域。进而,在梯度重建的基础上利用标记分水岭变换得到分割图像。其次,采取空间相邻关系、面积、光谱与纹理等多约束策略,以搜索最小合并代价的方式合并最初分割区域中的邻接区域对。最后,修改细节子图并进行小波逆变换将最初分割结果投影到更高尺度图像,同时处理边界上的像元以保持区域边界直至原始图像。实验结果表明本文方法不仅能够用于高分辨率多光谱遥感图像的分割,而且缓解了过分割问题且取得了较准确的分割效果。     

一种基于坐标改正量的大区域GIS数据库的坐标系统转换方法

随着计算机海量空间信息存储和处理能力的显著提高,我国基础地理信息数据组织已由传统的分幅方式转向了数据库方式,由此大区域GIS数据库坐标系统转换成为一个亟待解决的问题。本文基于实际生产中较易获取的分幅坐标改正量资料,提出了一种利用Delaunay三角网剖分的GIS数据库的坐标系统转换方法,对大区域内GIS数据库坐标系统分区逐点转换的原理、思路与流程进行了深入探讨,并开展区域性实验对该方法的效果进行了验证。     

基于拓扑约束的方向关系定性描述模型

针对原有方向关系矩阵模型对于参考目标MBR区域的方向描述缺陷问题,本文将拓扑约束引入方向关系定性描述,构建基于拓扑参考的方向关系定性描述模型,实现了MBR区域方向关系的有效表达。新模型首先将参考目标的MBR区域划分为不同的拓扑区域,提出方向关系拓扑参考定义;基于拓扑参考,分别对不同拓扑区域定义相应的方向关系矩阵;最后,根据参考目标与源目标间的不同拓扑关系,提出不同情况下方向关系分层定性描述策略。实验结果表明,新模型充分反映了拓扑关系对方向关系描述的约束关系,能有效提高方向关系表达的准确性和精确性。     

基于北斗LBS技术的新疆地理国情普查安全生产监控系统的设计研究

在荒漠或人烟稀少地区,需要先进的应急救援管理手段来保障野外测绘作业人员的生命安全,基于北斗卫星导航系统的LBS技术能够解决安全生产问题。本文在分析新疆地理国情普查安全生产监控系统建设总体结构的基础上,阐述了其设计方案和安装部署,并探讨了系统各功能的实现方法,并验证了该系统能够为野外作业人员的位置定位、报警及指挥中心的监控和应急救援发挥作用。     

Incorporating sprawl and adjacency measures in land‐use forecasting model: A case study of Collin County,TX

Sprawl measures have largely been neglected in land‐use forecasting models. The current approach for land‐use allocation using optimization mostly utilizes objective functions and constraints that are non‐spatial in nature. Application of spatial constraints could take care of the contiguity and compactness of land uses and can be utilized to address urban sprawl. Because a land‐use model is used as an input to transportation modeling, a better spatial allocation strategy for more compact land‐use projections will promote better transportation planning and sustainable development. This study formulates a scenario‐based approach to normative modeling of urban sprawl. In doing so, it seeks to improve the land‐use projections by employing a spatial optimization model with contiguity and compactness consideration. This study incorporates urban sprawl measures based on smart growth principles together with a mixed‐use factor, and adjacency consideration of nearby land uses. The objective function used in the study maximizes net suitability based on imposed constraints. These constraints are based on smart growth principles that enhance walkability in neighborhoods, promote better health for residents, and encourage mixed‐use development. The formulated model has been applied to Collin County, TX, a fast‐developing suburban county located to the north of the Dallas–Fort Worth metroplex. The suitability of land cells indicates the probability of conversion, which is calculated using spatial discrete choice analysis with Moran eigenvector spatial filtering for vacant cells at a resolution of 150 × 150 m employing factors of the built environment, and socioeconomic and demographic characteristics. This study demonstrates how spatial proximity between land uses, which has been ignored to date, can be used to control sprawl, resulting in better mixing of different land uses based on constraints imposed in a spatial optimization problem.     

Revealing the spatiotemporal characteristics of the general public's panic levels during the pandemic crisis in China

The existing crisis management research mostly reveals the patterns of the public's panic levels from the perspectives of public management, sociology, and psychology, only a few studies have revealed the spatiotemporal characteristics. Therefore, this study investigates the spatial distribution and temporal patterns and influencing factors on the general public's panic levels using the Baidu Index data from a geographic perspective. The results show that: (1) The public's panic levels were significantly correlated with the spatial distance between the epicenter and the region of investigation, and with the number of confirmed cases in different regions when the pandemic began to spread. (2) Based on the spatial distance between the epicenter and the region, the public's panic levels in different regions could be divided into three segments: core segment (0–500 km), buffer segment (500–1300 km), and peripheral segment (>1300 km). The panic levels of different people in the three segments were consistent with the Psychological Typhoon Eye Effect and the Ripple Effect can be detected in the buffer segment. (3) The public's panic levels were strongly correlated with whether the spread of the infectious disease crisis occurred and how long it lasted. It is suggested that crisis information management in the future needs to pay more attention to the spatial division of control measures. The type of crisis information released to the general public should depend on the spatial relationship associated with the place where the crisis breaks out.     

基于MapX的桂林市自然景观特色观景区基础创造功能的算法实现

实现桂林市自然景观特色观景区创造的首要问题是如何依据桂林市的"山"水"特色确定特色观景区的空间平面区域,即特色观景区的基础创造,本文通过算法设计解决了基于MapX实现特色观景区基础创造功能过程中产生的特色观景区重叠、特色观景区公共边及环形特色观景区等问题。该功能的实现为进一步实现特色观景区的创造奠定了参考基础。     

利用灰集进行空间区域拓扑关系不确定性定量分析

针对不确定性空间区域对象间拓扑关系不确定性的定量分析问题,提出了一种具有双隶属度函数的灰集表示模型,利用双隶属度函数的截集将不确定性空间区域对象转化为确定性区域对,借用相关的不确定性空间拓扑关系定性分析模型,将不确定性空间对象间拓扑关系的不确定性量化为一个可信度区间。     

Spatial representation in the social interaction potential metric: an analysis of scale and parameter sensitivity

The social interaction potential (SIP) metric measures urban structural constraints on social interaction opportunities of a metropolitan region based on the time geographic concept of joint accessibility. Previous implementations of the metric used an interaction surface based on census tracts and the locations of their centroids. This has been shown to be a shortcoming, as the metric strongly depends on the scale of the zoning system in the region, making it difficult to compare the SIP metric between metropolitan regions. This research explores the role of spatial representation in the SIP metric and identifies a suitable grid-based representation that allows for comparison between regions while retaining cost-effectiveness with respect to computational burden. We also report on findings from an extensive sensitivity analysis investigating the SIP metric’s input parameters such as a travel flow congestion factor and the length of the allowable time budget for social activities. The results provide new insights on the role of the modifiable areal unit problem in the computation of time geographic measures of accessibility.