Research On Methods And Integration Applications Of Polygonal Object Matching On Multi-scale Datasets

Recent trends in Geographic Information Science necessitate the integration of constantly expanding geospatial datasets.Integration provides an enriched product by improving positional or semantic accuracies and reduces costs between users.The first step in the integration process is to find corresponding objects and establish their linkages between two geospatial datasets,which is referred to as object matching.It also plays an important role in updating,evaluating,and managing spatial data.Therefore,object matching poses a fundamental research problem in geospatial datasets.This thesis studies selected the polygonal object matching methods.Based on these matching methods,this thesis further explores the approaches of improving the quality of crowdsourced geospatial information and spatial data integration between sea base topographic data and land base topographic data,and provides rich,accuracy,and current spatial data support for automation.The main research works of the dissertation are as follows:(1)A comprehensive review and summary of the polygonal object matching method is introduce.The object matching aim to find corresponding objects.This thesis first defines the concept of the corresponding objects from three aspects:apatial,attribute and time expression.Then,the source and from of the differences between the corresponding objects is presented.Finally,the most complex M:N matching are analyzed and the limitations that the current matching methods solve M:N matching are discussed.(2)The matching data are often obtained from different sources and have problems of positional discrepancy and different levels of detail(LoDs).To resolve these problems,this thesis presents a multiscale polygonal object-matching approach,called the minimum bounding rectangle combinatorial optimization(MBRCO)with spatial district(SD).This method starts with the MBRCO algorithm and its enhancement using the SD to find corresponding MBRs of one-to-one,one-to-many,and many-to-many matching pairs.Then,it aligns the MBRs of the matching pairs to identify object-matching pairs,which are evaluated using a matching criterion to find geometrically corresponding objects.(3)This thesis designs an iteration matching framework which effectively combines the advantages of the contextual information and artificial neural network(ANN).The proposed method can aggregates correct 1:N and M:N potential matching pairs using contextual information under positional discrepancy and a high spatial distribution density scenario and iteratively detects new landmark pairs according to the old landmark pairs until all the landmark pairs are obtained.(4)This thesis proposes a building polygon alignment approach for spatial data integration which can be used to improve the spatial data position accuracy.Firstly,this paper uses the MBR combinatorial optimization algorithm to identify corresponding objects between the integrated data.Then,a pairwise constraint spectrum matching algorithm based on geometric similarity is proposed to detect conjugate-point pairs of 1:1,1:N,and M:N correspondence.The conjugate-point pairs from 1:N and M:N correspondence inevitably have weak or error corresponding point pair,thus this paper proposes a least-squares algorithm based on the IGG1 weight to align the corresponding objects.The proposed method is applied to align base map data with higher positional accuracy and Google map data with lower positional accuracy.(5)In the Zhejiang DLG database,1:2000 island base topographic data,1:10000 mud flat base topographic data,1:25000 underwater topographic data,and 1:10000 land base topographic data are independent and they exist overlapping area.In this thesis,the polygonal object in sea base topographic data and land base topographic data are intergrated base on the proposed matching methods.We aim to achieve the accuracy improvement,redundancy reduction and unified management of sea base topographic data and land base topographic data.Firstly,establish a unified coding and integration library.Then,establish the linkage between the corresponding objects through the proposed matching method.Finally,intergrate them based on the principle.