面向地表覆盖变化检测的服务关系模型与方法研究

发布时间：2018-07-03 00:34

本文选题：地表覆盖更新 + 在线变化检测　；参考：《中国矿业大学(北京)》2017年博士论文

【摘要】：遥感影像变化检测是大范围地表覆盖数据更新的重要技术手段。近年来,国内外研究学者从不同的角度提出了大量的变化检测算法或模型,但尚未有一种通用的方法能够适用于所有的影像数据条件、地表覆盖数据类型和地理区域特点。因此,在应对大范围地表覆盖变化监测和更新任务时,用户需要根据专家经验知识选择较优的变化检测算法,继而构建处理流程。事实上,变化检测领域的算法与数据之间、数据与数据之间以及算法与算法之间存在着诸如竞争、协作、约束等关联关系,深入理解这些"算法-数据"关系将有助于实现变化检测算法的优选以及处理流程的灵活构建。然而,目前国内外对变化检测"算法-数据"关系的系统性总结、形式化描述和表达以及应用等方面均缺乏深入的研究。当前的一个重要发展趋势是借鉴Web Service、服务计算等理论方法,将变化检测所涉及的数据和算法以Web Service的方式进行封装与发布,发展基于服务计算的地表覆盖在线变化检测。本文在Web Service的背景下,将变化检测的"算法-数据"关系映射为Web服务关系,研究了服务关系建模、表达、推理等理论,提出了变化检测服务关系的分类与描述方法、构建了变化检测服务关系的表达模型、发展了基于服务关系的变化检测处理动态构建方法,主要研究内容与取得的成果如下:(1)研究了变化检测“算法-数据”关系映射为Web服务关系的问题,提出了变化检测服务关系的分类与描述方法。以山东省为实验区域,设计了大范围地表覆盖变化检测实验,研究了变化检测算法选择、处理流程定制等面临的技术问题。剖析了地表覆盖变化检测中存在的"算法-数据"关系及其封装为Web服务时映射形成的Web服务关系,进而提出基于服务语义的变化检测服务关系分类方法,即通过分析变化检测服务的数据、操作、执行和QoS语义,凝练出"接口-行为-执行"三个层级9种类型的Web服务关系,其中接口层3种,行为层4种,执行层2种。最后,综合考虑变化检测领域数据的属性特征、处理服务的多级粒度以及服务接口语义匹配关系等因素,利用描述逻辑(Description Logic)对这9种的服务关系进行形式化的定义与描述。(2)研究了服务关系组织、表达、存储等问题,构建了地表覆盖变化检测服务关系的表达模型。首先,在上述服务关系分类和描述的基础上,明晰了服务、服务语义、服务关系三类对象元素及其之间的聚合、依赖等关联关系,建立了基于ORD(Object and Relation Diagram)的服务关系概念模型。然后,借鉴超图模型在多元关系表达中的优势,定义了变化检测服务关系超图(Service Relation hyper Graph for Change Detection,SRhG-CD),提出了基于超图的逻辑表达模型,利用简单边、超边、有向边和无向边有效地表达了服务间诸如竞争关系、顺序关系、隶属关系等多粒度、多元性的服务关系,突破了传统的有向图模型在表达多元服务关系时的回路问题,避免了其在推理过程中引发的死循环现象。最后,提出了基于OWL DL的变化检测服务关系形式化表达与物理存储策略,发展了一种基于语义标注和服务关系判定的SRhG-CD模型更新演化方法,为服务关系的推理和应用奠定了基础。(3)研究了变化检测服务链按需生成的问题,发展了基于服务关系的变化检测处理动态构建方法。依据提炼的"接口-行为-执行"三层服务关系,构建了用于确定变化信息提取服务和数据预处理服务的4种推理规则,采用SWRL(Semantic Web Rule Language)对推理规则进行形式化表达,进而构建了基于规则多类别语义匹配的闭包服务集,缩小了服务链按需生成时的服务查找空间。并在此基础上,将用户需求描述为起始数据、期望数据和期望QoS的三元组模型,以用户起始数据和期望数据为功能性驱动,提出了一种基于后继服务查找的服务链生成算法。以用户期望QoS为非功能性驱动,发展了一种变化检测服务链评价方法,实现了服务链的最优化选取。该变化检测处理动态构建方法解决了传统的服务链固定编排调用带来的处理构建自动化程度低的问题,实现了用户功能性和非功能性需求约束下变化检测服务链的灵活生成。(4)研制了地表覆盖在线变化检测实验原型系统在上述理论研究的基础上,基于服务计算、WebGIS等技术研制了地表覆盖在线变化检测实验原型系统(GLC-OCD)。基于ArcGIS Server、WSDL、OGC WPS三种常用的服务标准,给出了变化检测处理服务封装与发布的过程。设计了基于Protégé软件的服务关系建模、SRhG-CD模型更新演化实验。在GLC-OCD平台上,给出了地表覆盖在线变化检测的具体案例,依据用户提供的初始数据(2010年的Landsat 5 TM和2015年的Landsat 8 OLI数据)和期望数据(变化区域和变化类别数据),系统动态地构建出8条满足用户功能性需求的服务链;依据用户提供的期望QoS("处理时间","小于","22","分钟"),系统推荐了1条最优的服务链。实验结果表明,在变化检测服务关系模型与处理动态构建方法的支持下,借助于GLC-OCD系统,可实现用户需求驱动下的变化检测服务链的灵活构建与在线执行,为地表覆盖大范围变化监测与更新提供了新思路新方法。
[Abstract]:Remote sensing image change detection is an important technical means for large scale surface coverage data updating. In recent years, researchers at home and abroad have proposed a large number of change detection algorithms or models from different angles, but there is not a general method that can be applied to all image data items, and the surface coverage of data types and geographical features. Therefore, in response to large scale surface coverage change monitoring and updating tasks, users need to select better change detection algorithms based on expert experience knowledge and build processing processes. In fact, there are competition, collaboration, and constraints between algorithms and data in the field of change detection, between data and data, and between algorithms and algorithms. Understanding these "algorithmic data" relationships will help to realize the optimization of change detection algorithms and the flexible construction of processing processes. However, the systematic summary of the "algorithm data" relationship, formal description, expression and application of the change detection "algorithm data" are lack of in-depth research at present. The development trend is to use the Web Service, service computing and other theoretical methods to encapsulate and publish the data and algorithms involved in the change detection in the way of Web Service, and develop the online change detection of surface coverage based on service computing. In this paper, the "algorithm data" relationship of change detection is mapped to Web service in the context of Web Service Relations, study the theory of service relation modeling, expression, reasoning and so on, propose the classification and description method of change detection service relation, construct the expression model of change detection service relation, develop the dynamic construction method of change detection processing based on service relation, the main research content and the achievements are as follows: (1) studying the change detection The relationship between algorithm and data is mapped to Web service relation, and the classification and description method of change detection service relationship is proposed. Taking Shandong Province as the experimental area, a large range surface cover change detection experiment is designed. The technical problems faced by the selection of change detection algorithm and process customization are studied. The change of surface cover change is analyzed. The "algorithm data" relationship in the detection and the Web service relationship that the package is mapped to Web service, and then the classification method of the service relation based on the service semantics is proposed, that is, by analyzing the data, operation, execution and QoS semantics of the change detection service, 9 types of Web suits are trained by the three levels of "interface behavior execution". There are 3 kinds of interface layer, 4 kinds of behavior layer and 2 kinds of execution layer. Finally, we consider the attribute characteristics of the data in the change detection field, deal with the multilevel granularity of the service and the semantic matching relation of the service interface, and use the description logic (Description Logic) to formalize the definition and description of the 9 kinds of service relations. (2) research (2) Service relations organization, expression, storage and other problems, construction of the expression model of surface cover change detection service relationship. First, on the basis of the classification and description of the above service relations, the service, service semantics, service relations and the three kinds of object elements and the relation between the aggregation and dependence are defined, and the ORD (Object and Relatio) is established. N Diagram) concept model of service relations. Then, drawing on the advantages of hypergraph model in the expression of multivariate relations, we define the change detection service relation hypergraph (Service Relation hyper Graph for Change Detection, SRhG-CD), and propose a logical expression model based on hypergraph, which is effectively expressed by simple edges, super edges, both side and undirected edges. The service relations, such as competition relation, sequence relationship and subordinate relationship, break through the loop problem of the traditional directed graph model in the expression of multiple service relations and avoid the dead cycle phenomenon in the reasoning process. Finally, the formal expression of the change detection service relationship based on OWL DL is proposed. With the physical storage strategy, a new method of updating and evolving SRhG-CD model based on semantic annotation and service relation decision is developed, which lays the foundation for the reasoning and application of service relations. (3) the problem of changing detection service chain on demand is studied, and the dynamic construction method of change detection processing based on service relationship is developed. The interface - behavior - executes "three layers of service relations, constructs 4 reasoning rules for determining the change information extraction service and data preprocessing service, and uses SWRL (Semantic Web Rule Language) to formally express the reasoning rules, and then constructs a closed packet service set based on the rules and multiple category semantic matching, which reduces the service chain in demand. On this basis, the user requirements are described as the starting data, and the three tuple model of the expected data and expected QoS is expected to be driven by the user initial data and expected data. A service chain generation algorithm based on the successor service lookup is proposed. The user expected QoS is a non functional driver and develops a non functional drive. The change detection service chain evaluation method has realized the optimal selection of service chain. The dynamic construction method of the change detection and processing solves the problem of low automation in the processing of the traditional service chain fixed arrangement call, and realizes the flexible generation of the change detection service chain under the requirements of the user's functional and non functional requirements. 4) on the basis of the above theoretical research, based on the above theoretical research, the prototype system of surface coverage on-line change detection (GLC-OCD) is developed on the basis of the above theoretical research. Based on the three common service standards of ArcGIS Server, WSDL and OGC WPS, the package and hair of the change detection and processing service are given. A service relationship modeling based on Prot e g is designed, and the SRhG-CD model updating and evolution experiment is designed. On the GLC-OCD platform, a specific case of the on-line change detection of surface coverage is given, based on the initial data provided by the user (the Landsat 5 TM in 2010 and the Landsat 8 OLI data in 2015) and the expected data (changing regions and changes) The system dynamically builds 8 service chains to meet the user's functional requirements; according to the expected QoS ("processing time", "less", "22", "minute") provided by the user, the system recommends 1 optimal service chains. The experimental results show that, with the support of the change detection service relation model and the processing dynamic construction method, the system has the aid of GLC-OCD The system can realize the flexible construction and online execution of the change detection service chain driven by the user needs, and provide a new method for monitoring and updating the large scale change of the surface.
【学位授予单位】：中国矿业大学(北京)
【学位级别】：博士
【学位授予年份】：2017
【分类号】：P237

【参考文献】