KMS XINJIANG INSTITUTE OF ECOLOGY AND GEOGRAPHY，CAS
| 城市空间格局多源遥感协同提取与特征分析 | |
| Alternative Title | Multi-source remote sensing collaborative extraction and feature analysis of urban spatial pattern |
| 李治 | |
| Subtype | 博士 |
| Thesis Advisor | 周成虎 ; 陈曦 |
| 2019-06-30 | |
| Degree Grantor | 中国科学院大学 |
| Place of Conferral | 北京 |
| Degree Discipline | 理学博士 |
| Keyword | 高空间分辨率遥感 多源空间信息协同 多特征提取 城市空间格局分析 High-spatial-resolution Remote Sensing Multi-source Spatial Information Collaboration Multi-feature Extraction Urban Spatial Pattern Analysis |
| Abstract | 城市景观组分及其空间格局研究是城市生态学中的重要研究方向之一，其对于城市的功能区优化，生态保护，规划管理和可持续发展等方面具有重要的科学意义与应用价值。高空间分辨率遥感影像（高分遥感影像） 具备精度高、大区域和重放周期短等对地观测优势，能够为客观的城市景观组分监测和科学的空间格局特征分析提供精细的空间信息。 然而， 由于高分影像的信息缺失和城市空间格局的复杂性对城市景观组分高分遥感信息提取和空间格局分析带来诸多挑战。 因此，城市景观组分的遥感高维特征挖掘与多源信息协同提取以及城市空间格局多层级联合分析是城市景观组分信息提取与空间格局分析中亟待解决的重要科学问题。本研究以遥感信息图谱理论和城市景观生态学为理论指导，基于多源信息协同的高分遥感数据， 从人文和自然视角出发，提出了城市人工景观高维特征挖掘和城市自然景观多源特征协同的提取方法。在此基础上， 发展了多层级联合的城市空间格局分析方法， 并以中国九个城市开展应用示范。 结果表明， 本研究方法在城市景观组分提取及空间格局特征分析中，均取得较优的效果，为城市的可持续发展提供方法支撑与信息服务。论文主要研究内容与结论包括：1） 揭示了高分遥感影像形态特征与城市人工景观要素（建筑物和道路）之间的有机联系，面向城市建筑物和道路景观要素提出了高维形态学特征表达方法。 城市建筑物和道路在高分影像中具有显著的空间形态特征， 而数学形态学在图像形态表达中具有显著的优势，且具备理论基础扎实和可扩展性好的特点， 因此， 可以作为重要的技术方法载体进行改进。 基于此， 在城市建筑物方面， 本研究将“自下而上”的建筑物多方向-尺度的结构表达和“自上而下”的知识准则约束联合， 提出了多尺度结构-准则联合的形态特征表达方法对城市建筑物进行高维表达； 在道路方面， 顾及道路的长度、宽度和方向的形态特征信息，提出改进的差分形态轮廓（MDMP）的多尺度形态特征。 在此基础上， 引入多源专题信息产品（GUF 和 GHSL 建筑物和 OSM 道路产品）， 发展面向面状和线状样本的自适应分割算法， 自适应的获取城市建筑物和道路组分信息。 实验结果表明， MDMP 特征和多尺度结构-准则联合的形态特征有效的增强了高分遥感中的城市道路和建筑物的的可表达性。2）构建了面向城市自然景观要素（植被和水体）的多源遥感特征协同表达方法，形成了本文的多尺度谱-空间形态协同的城市地表水提取方法和多尺度纹理-时间序列协同的城市植被提取方法。 面向城市自然景观组分（城市地表水和植被要素），针对高分遥感数据的光谱信息匮乏及难于获取高时间维度信息的问题， 协同 Landsat中分数据的光谱信息和高时间维信息，形成了多尺度光谱-空间形态特征城市地表水表达模型和时间谱-纹理特征城市植被表达模型。实验结果表明， 本研究的城市水体和植被信息的提取精度均有提升，且在不同的实验场景下表现出良好的提取效果，表明本研究所提出的协同方法具有较好的鲁棒性和泛化能力。3） 构建建成区-街区-斑块的多层级基底嵌套结构， 借助景观格局指数和形状指数，发展面向多层级联合的评价指标与分析方法， 开展中国九个城市（北京，上海，广州，武汉，成都，西安，沈阳，乌鲁木齐和拉萨） 的应用示范，挖掘典型城市间的空间格局典型特征， 反映城市化和生态现状，为合理的土地利用规划和管理，城市生态与环境建设提供信息支撑。 应用景观生态学的基底-廊道-斑块理论， 构建城市建成区-街区-要素的多层次结构地理单元， 综合运用景观格局指数、形状分析指数、景观组分等评价方法， 发展了多层级联合的城市基底-廊道-斑块分析方法， 挖掘与分析九个城市空间格局特征，揭示其城市化、城市绿化等空间格局模式。 结果表明：本文提出的多层级城市空间格局分析方法，具有实际的物理意义和应用价值，避免了单一尺度分析或者同质多尺度分析带来的局限性，提升了对城市空间格局的多视角认知，可以更为客观和精细的反映城市化和城市生态现状。本文针对高分遥感影像缺失信息的特征提取和单层级城市空间格局分析问题，在遥感信息图谱理论和景观生态学框架指导下， 按照城市的“基底—廊道—斑块景观组分提取和空间格局特征分析”的主线开展研究，对围绕高分遥感的多源特征挖掘与协同提取及城市空间格局分析的理论方法和技术难点进行了深入研究，为城市高分遥感的空间格局提取和分析的难题提供了一个新的解决途径。 |
| Other Abstract | The study of urban landscape components and their spatial pattern is one of theimportant research directions in urban ecology. It has important scientific significance andapplication value for urban functional area optimization, ecological protection, planningmanagement and sustainable development. The high spatial resolution remote sensingimage (high-resolution remote sensing image) has the advantages of high-precision,large-area and high-time ground observation, and can provide fine spatial information forobjective monitoring of urban landscape components and scientific analysis of spatialpattern characteristics. However, due to the limitations of high-resolution images and thecomplexity of urban spatial patterns, there are many challenges to the high-resolutionremote sensing information extraction and spatial pattern analysis of urban landscapecomponents. Therefore, remote sensing high-dimensional feature mining and multi-sourceinformation collaborative extraction of urban landscape components and multi-level jointanalysis of urban spatial patterns are important scientific issues to be solved in urbanlandscape component information extraction and spatial pattern analysis. Based on thetheory of remote sensing information atlas and urban landscape ecology, this paperproposes a high-dimensional feature mining method for urban artificial landscape elementsand multi-source features for urban natural landscape elements based on multi-sourceinformation collaborative high-scoring remote sensing data. Collaborative adaptiveextraction method. On this basis, a multi-level joint urban spatial pattern analysis methodwas developed, and application demonstrations were carried out in nine cities in China.The results show that the research method has achieved superior results in urban landscapecomponent extraction and spatial pattern characteristics analysis, and provides methodsupport and information services for urban sustainable development. The main researchcontents and conclusions of the thesis include:1) It reveals the organic connection between the morphological features ofhigh-resolution remote sensing images and urban artificial landscape elements(buildings and roads), and proposes multi-scale morphological feature extractionmethods for urban buildings and road landscape elements. Urban buildings and roadshave significant spatial morphological features in high-resolution images, whilemathematical morphology has significant advantages in image morphological expression,and has a solid theoretical foundation and good scalability. Therefore, it can be regarded asimportant. The technical method carrier is improved. Based on this, in terms of urbanbuildings, this study combines the multi-directional-scale structural expression of“bottom-up” buildings with the “top-down” knowledge criteria constraints, and proposes multi-scale structure-criteria joint The morphological feature representation method is usedto express the high-dimensional representation of urban buildings. On the road,considering the morphological characteristics of the length, width and direction of the road,the improved multi-scale morphological features of the differential morphology contour(MDMP) are proposed. Introduce multi-source thematic information products (GUF andGHSL buildings and OSM road products), develop adaptive segmentation algorithms forplanar and linear samples, and adaptively acquire information on urban buildings and roadcomponents. The experimental results show that the combined morphological features ofMDMP features and multi-scale structure-criteria effectively enhance the expressibility ofurban roads and buildings in high-resolution remote sensing.2) Constructing a multi-source remote sensing feature collaborative expressionmethod for urban natural landscape elements (vegetation and water body), forming amulti-scale spectral-spatial morphological collaborative urban surface waterextraction method and multi-scale texture-time series collaborative city. Vegetationextraction method. Facing urban natural landscape components (urban surface water andvegetation elements), the lack of spectral information for high-scoring remote sensing dataand the difficulty of obtaining high time-dimensional information, and the spectralinformation and high time-dimensional information of Landsat's sub-data are formed.Multiscale Spectral-Spatial Morphological Characteristics Urban Surface Water ExpressionModel and Time Spectrum-Texture Characteristics Urban Vegetation Expression Model.The experimental results show that the extraction accuracy of urban water and vegetationinformation in this study is improved, and it shows good extraction effect under differentexperimental scenarios, which indicates that the synergistic method proposed in this studyhas better robustness and generalization.3) Constructing a multi-level base nesting structure of built-uparea-street-plaque, using the landscape pattern index and shape index to developevaluation indicators and analysis methods for multi-level joint development, andlaunching nine cities in China (Beijing, Shanghai, Guangzhou, Applicationdemonstrations in Wuhan, Chengdu, Xi'an, Shenyang, Urumqi and Lhasa to explorethe typical characteristics of spatial patterns between typical cities, reflectingurbanization and ecological status, providing information support for rational landuse planning and management, urban ecology and environmental construction.Applying the base-cave-plaque theory of landscape ecology, constructing a multi-level structural geographic unit of urban built-up area-street-element, comprehensively applyingthe evaluation method of landscape pattern index, shape analysis index and landscapecomposition, and developing multi-level The combined urban basement-corridor-plaqueanalysis method mines and analyzes the spatial pattern characteristics of nine cities andreveals the spatial pattern patterns such as urbanization and urban greening. The resultsshow that the multi-level urban spatial pattern analysis method proposed in this paper haspractical physical meaning and application value, avoiding the limitations brought bysingle-scale analysis or homogeneous multi-scale analysis, and enhances themulti-perspective recognition of urban spatial pattern. It is known that the status quo ofurbanization and urban ecology can be more objectively and carefully reflected.In this paper, based on the characteristics of high-resolution remote sensing imagemissing information and homogenous urban spatial pattern analysis, under the guidance ofremote sensing information atlas theory and landscape ecology framework, according tothe city's “basal-corridor-plaque landscape component extraction and The main line ofspatial pattern analysis is carried out to study the theoretical methods and technicaldifficulties of multi-source feature mining and collaborative extraction and urban spatialpattern analysis around high-resolution remote sensing, and to extract and analyze thespatial pattern of urban high-scoring remote sensing. |
| Subject Area | 地图学与地理信息系统 |
| Language | 中文 |
| Document Type | 学位论文 |
| Identifier | http://ir.xjlas.org/handle/365004/15339 |
| Collection | 中国科学院新疆生态与地理研究所 研究系统 |
| Affiliation | 中国科学院新疆生态与地理研究所 |
| First Author Affilication | 中国科学院新疆生态与地理研究所 |
| Recommended Citation GB/T 7714 | 李治. 城市空间格局多源遥感协同提取与特征分析[D]. 北京. 中国科学院大学,2019. |
| Files in This Item: | There are no files associated with this item. | |||||
Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.
Edit Comment