Research On Methods Of Topology Parallel Checking Of Cadastral Database Based On CUDA

Cadastral database is one of spatial database,which involves the land rights elements are:boundary point,boundary line and parcel.The topology checking based on specific rules is the key content of cadastral data quality control.With the advent of the big data era,due to the increasing and complexity of the cadastral data,the method of spatial topological check in traditional serial mode is far from satisfying the requirement in practice.In consideration of the complex topological checking algorithm computation of traditional cadastral and the inefficient serial mode,a method for topology parallel checking of cadastral database based on CUDA,which is practical,efficient and adapted to the new generation of computer GPU hardware architecture,is proposed.There are several steps in the method system.Firstly,in order to getting the processed object combination,the spatial data should be filtered.Then based on the description of the rules constraint of cadastral database topology relation,three kinds of atomic operation which are concluded from the basic topological atomic rules are proposed.And a parallel algorithm for topology parallel checking between cadastral objects based on CUDA is designed and implemented.By introduced the parallel algorithm into the application of cadastral database topology checking,the whole process of the cadastral topology checking is presented.The major work implemented and conclution are presented as follows:(1)researching on methods of parallel spatial index.The efficiency of spatial data processing depends on the performance of spatial indexing technology.In this paper,the research results of the existing spatial index are summarized.And the construction ideas suitable for CUDA parallel architecture of the parallel index are porposed,which provides the basis for data filtering.(2)researching on methods of the cadastral topological relation based on CUDA.Considering the existing topology basic algorithm is not sufficient,the process of topology algorithm is designed again.Therefore a Parallel computing model based on CPU and GPU is presented.In the model,the GPU is responsible for handling the computation-intensive tasks and the CPU assumes the situation judgment and task scheduling work.And through the experiment,it can be verified that the algorithm can achieving high performance computing tasks in a fast and efficient manner.(3)Construction of the prototype system and verification of cadastral topology algorithm.The prototype checking system of cadastral database is constructed by language C++ and CUDA.By this system,the parallel topology-checking method proposed in this paper is tested under different data sizes.And the experimental results are compared with the common commercial software(for example ARCGIS).Finally,the accuracy and efficiency of the experimental results are evaluated.And through the experiment,it can be verified that the methods of topology parallel checking of cadastral database based on CUDA is feasible,accurate and efficient.