ARTICLES
XU Feng, DENG Min, ZHAO Binbin, CHEN Jianjun
2009, 11(5): 657-663.
To improve both the reuse and the quality of existing spatial data,it is urged to develop approaches for integrating and updating spatial data,which possibly are collected from multiple sources,at different time,and/or multiple representations for the same spatial region.For this purpose,object matching has become one of the crucial technologies,and it is conducive to detect change and/or difference between different spatial datasets(e.g.two different scales of maps)by comparison of corresponding objects.Currently many matching methods have been developed for different types of objects in practical applications.These methods can be classified by two modes,i.e.matching criterion and object type.Methods based on matching criterion can be classified into three categories,i.e.geometric,topological and semantic matching,which are respectively implemented by geometric constraints(e.g.distance,shape similarity,direction),topological constraints(e.g.connectivity,node degree) and semantic constraints(e.g.semantic similarity),whereas the matching methods based on object type can be further referred as point-to-point,point-to-region,line-to-line,line-to-region,and region-to-region matching.In the process of object matching,there are possibly six matching mapping relations,i.e.1∶0,0∶1,1∶1,1∶N,N∶1,N∶M.As for the latter three matching mapping,an iterative searching process is needed to build the mapping relations between the correspondences in two different spatial datasets.Through analysis of literatures,one can find that each of the matching methods to some extent has its advantages for some particular types of objects(e.g.line or area) and/or applications.That is to say,they are lack of generality.Based upon this situation,this paper makes a systematical investigation on existing matching methods,including their classifications,basic principles and procedures,and the applications in geo-spatial information science(e.g.map updting,spatial data integration,etc).A detailed comparison is made for these matching methods from aspects of matching criterion,object type and scale.In particular,it is pointed out the characteristics of object matching with different scales of map updating,which is distinctly different from the object matching with same or similar scales.This work is very helpful for further development of general matching methods.