Modeling And Methods Of Service Relation For Land Cover Change Detection

Change detection using remote sensing imagery is one of the critical steps for land cover monitoring and updating in large area.During the last twenty years,a variety of change detection algorithms and techniques have been developed.Many of them worked well in experimental studies or smaller areas.However,none of them can be universally applicable for all kinds of imagery,different land cover types and geographic regions.One of its consequences is that the selection of change detection algorithms and the construction of processing workflows depend largely on the expertise of experts or experienced operators.In change detection domain,there have relations exist between algorithm and data,between algorithms,and between datasets,which are referred in the following discussion as "algorithm-data" relations.A better understanding and representation of these "algorithm-data" relations can facilitate the optimum selection of change detection algorithm(s)and improve the intelligence of land cover change mapping or monitoring.However,a formal description and representation of these "algorithm-data" relations have not been fully discussed.Recently,it is an important trend to develop web-based land cover change detection with publishing the remote sensed data and processing algorithms or models into Web Services.In the context of Web Service,the "algorithm-data" relations in change detection are mapped into Web service relations.We study the theories of service relation modeling,representation and reasoning,propose a method of service relation classification and description,build a service relation representation model,and develop a method of service relation based dynamic construction of change detection workflow.The main researches include the following four parts.(1)Studying the process of “algorithm-data” relations mapping into web service relations,and proposing a method of service relation classification and descriptionTaking Shandong province as experimental area,an experiment is designed for land cover change detection in large area.The technical problems are researched including change detection algorithm selection and workflow construction.The mapping relationship between the "algorithm-data" relation and the web service relation is analyzed.Then,a service relation classification method is proposed based on service semantics.The "algorithm-data" relations are mapped into a set of web service relations with the analysis of functional and non-functional service semantics.These service relations are further classified into three different levels and nine categories,i.e.,interface,behavior and executor levels.Finally,description logic is used for formal description of these service relations.(2)Studying the organization,representation and storage of service relation,and building a service relation representation model for land cover change detectionOn the basis of the classification and description of service relation,a conceptual model is built by using ORD(Object and Relation Diagram)for clarifying the service,service semantics,service relations,and their incidence relations.Then,the definition of service relation hyper graph for change detection(SRhG-CD)is given by extending the traditional hyper graph.SRhG-CD can be used for representation the multi-granularities and multivariate service relations,which avoid the limitation of the traditional methods only be applied for the binary service relations.Finally,OWL DL is used for the formal description and storage of SRhG-CD.A updating strategy for SRhG-CD model is discussed and proposed based on semantic annotation and service relation mining.(3)Studying the problem of on-demand service chain generation,and developing a method of dynamic construction of change detection workflow based on service relationA set of reasoning rules are firstly built based on the "interface-behavior-executor" service relations,which are formal described using SWRL(Semantic Web Rule Language).Then,a SWRL matching based closure service set is constructed for improving the efficiency of service finding.On this basis,the users requirements are described with three tuple of start data,expect data and expect QoS.An on-demand service chain generation method is proposed based on sequential service finding.Finally,a service chain evaluation and optimization strategy for change detection is proposed.The change detection service chain can be flexible generated under the constraint of user’s functional and non-functional requirements,which solves the difficult problem of fixed service chain of the traditional method.(4)Implementing a web-based system for online land cover change detectionOn the basis of the above theories,a web-based system for online land cover change detection(GLC-OCD)is designed and developed using service computing and WebGIS technologies.Three service encapusation and publish methods are given based on ArcGIS Server,WSDL and OGC WPS.The experiments of service relation modeling and SRhG-CD model updating are designed and realized.With the GLC-OCD,a case study of web-based land cover change detection is given.On the basis of user’s start data(Landsat 5 TM in 2010 and Landsat 8 OLI in 2015)and expect data(change area and change type data),8 change detection service chains are built for meeting user’s functional requirements.On the basis of user’s expect QoS("execute time","less than","22","min"),one optimal service chain is recommended.The results indicate that with the help of change detection service relation model and workflow dynamic construction method,on-demand service chain can be dynamically generated according to users’ requirements on the GLC-OCD,which provide a new solution for supporting land cover monitoring and updating in large area.