Road Network Matching Based On Stable Hierarchical Spatial Morphology

Vector map space objects matching is the core and key technology in the applications of spatial data integration and fusion,multi-scale spatial database maintenance and updating,location-based services and so on.Being applied to the integration and fusion of multi-source spatial data could reduce the high costs of data collection and help to improve the quality of spatial data.Applying it to the incremental updating of spatial data can greatly improve the efficiency of updating and reduce maintenance costs,which is the current hot issue for domestic and foreign scholars.Vector space objects matching technology of multi-source and multi-scale road network can not only serve the vector road map updating,integration,integration and other application,but also meet the rapid development of car navigation products' strong demand for navigation electronic map data updating.But with the complexity of multiple sources and multiple-scale space target matching relation that the road network reflected,the space target's matching method of road network has met various degrees of bottleneck in the precision and accuracy:1)Detection and processing of complex matching relations.2)Focus on the local shape of space targets in excess.3)Large scale data matching is inefficient.There is an urgent need to further develop the spatial information's similarity quantitative measure theory and match method of vector map.Aiming at the problem of the automatic matching of the vector space object in the road network,putting the point at the bottleneck of the accuracy and efficiency of the current matching method,this research will start our work from quantitative measurement method of road network's spatial information,fully integrate spatial cognition,extract stability and hierarchical spatial form as a stable reference,which is based on visual attention in spatial cognition mechanism,establish the similarity measure model between the road network entities with stability hierarchical spatial form.Meanwhile,in view of multi-source and multi-scale road network entity matching's complexity and uncertainty,we introduce intelligent optimization method and deal with road network entity matching,which according to similarity measure model that this research established,and combined the Particle Swarm Optimization method of intelligent optimization.We hope to form a new set of road network matching theories and methods which is an intelligent optimization theory-driven and based on the stable hierarchical spatial form.In order to improve the efficiency of matching,this paper introduces the idea of parallel acceleration to construct a parallel framework of MPI+OpenMP+CUDA(MOCP),and parallel transforms intelligent optimization matching algorithm based on this framework,finally realizes a parallel intelligent optimization of road network matching algorithm.Compared with the traditional matching algorithm,the parallel intelligent optimal matching algorithm has a great improvement in the accuracy and efficiency.This research also promotes the quantitative measurement theory of vector map spatial information and better serves the integrated analysis and application of multi-source and multi-scale road network vector data.These studies are divided into the following four aspects:1)Extraction of road network's stability of hierarchical spatial form.The human visual attention processes are from the primary to the secondary,from the simple to the complex,and gradually developing,this abstract description of the cognitive law is hierarchy.Then,this paper will simulate human cognitive mechanism of visual hierarchy to precede spatial hierarchical road network classification and inference,and extract road network level space form.In this way,we can start from the high level road of the salient region when matching,and use this as the basis to diffuse the neighborhood to a low level and then determine the matching relationship,it makes the matching process in line with the rules of human spatial cognition,and the hierarchical information can also provide hierarchical constraints for the matching between the road entities,thus improving the accuracy of matching.Mainly due to the incompleteness of the spatial entities' semantic information,multi-scale road entities excessively depend on local structure similarity measure method,which can be also regarded as the lack of semantic references.Therefore,this paper extracts the relatively stable reference form representing the stable form of urban space,and establishes the stable reference of other forms in addition to semantic reference.Based on the stability of space form,transforming the geometric similarity calculation between complex spatial objects into the matching of spatial objects in the same stable computing space can make the matching method in this paper not only stability but also global.2)Similarity measurement model of road entity based on stable hierarchical spatial morphology.The traditional matching methods mainly depend on the similarity measurement of the local morphological structure(the metrics are: distance,shape,direction,topology,etc.),which is easy to fall into local optimization.And,space form stable level provided a reference to global stability for road matching between entities.We can combine the stable reference similarity comparison;transform the two entities that which is ready to be compared into the same computing space to make the relative distance between the road entities and the stable space form to be the important characteristics of similarity measure index.The similarity measure model based on stable level space form has both global features and local characteristics;it can overcome the geometry structure between relying on local entities as similarity judgment according to the problems and improve the accuracy of matching between entities.3)Establishment of optimal matching algorithm for road network based on intelligent optimization.The development of intelligent optimization theory provides a new solution for the determination of complex matching relationship.The intelligent optimization method does not require the continuity and convexity of the objective function and constraint,but it has a strong ability to adapt to the uncertainty of the data,and the algorithm converges quickly.Multi scale road network of many complex situations,there is a match,the matching relationship can be understood as the uncertain features,and road network entity optimal matching solution coincides with the theory of intelligent optimization expression ability.In addition,in order to find the best matching result from the large scale candidate matching set,we can select the intelligent optimization search algorithm to achieve the global optimization in order to get the final result.Therefore,this paper will combine the stability level similarity measurement model with the particle swarm optimization algorithm and deal with the relationship between road networks by using the method of swarm intelligence optimization;it can provide a new theoretical and technical support for the complex matching of multi-source and multi-scale road network.The experimental data show that the road network matching method based on intelligent optimization has a great improvement in accuracy and efficiency.4)Parallel intelligent optimization of road network matching method.In order to solve the problem of low efficiency of the traditional matching algorithm for the large scale data of road network,the idea of parallel acceleration is introduced to the road network matching.This paper implements the parallel framework of MOCP,and combines the advantages of cluster,CPU and GPU respectively,so as to give full play to the computing ability of existing nodes and maximize the parallel computing ability.Based on the MOCP framework,the intelligent optimization of road network matching method is parallelized,and finally a multi-level parallel intelligent optimization of road network matching method has been realized.The experimental results show that the intelligent optimization algorithm based on the MOCP framework can improve the matching efficiency in the case of a large number of particles.In this paper,the data of several typical regions are selected from the multi-source and multi-scale road network in Wuhan,and we use the stable hierarchical spatial morphology similarity measure model and intelligent optimization method to carry out matching experiments of multi scale road network.We will measure the correctness and reliability of the new method proposed in this study by experimental cases.Meanwhile,we'll analyze the problem and improve the stable hierarchical spatial morphology framework,the similarity measurement model based on stable urban spatial morphology and the optimization matching method based on intelligent optimization theory developed in this paper,so as to further improve the theoretical research results.Through the research of this paper can further reveal the cognitive field,theoretical value and scientific significance in road network space,and at the same time,it provides a new theory and solution for the automatic matching of multi-scale spatial objects,which has important practical significance.