Study On The Optimization Of Large-scale Unmanned Aerial Vehicle Remote Sensing Data Processing Based On Cloud Computing

The application of unmanned aerial vehicle(UAV)remote sensing technology has been widely used in the fields of resource investigation and environmental monitoring.The analysis and processing of a large number of remote sensing images are involved in the application of this kind of remote sensing technology.In this scenario,it is difficult to store and use these date at present.Cloud computing and GIS technology are applied in the field of the processing of remote sensing images,which provide a new idea for remote sensing image storage and management.Based on Hadoop cloud computing architecture and GIS technology,this thesis designs the process from the generation of tile data to the storage of tile data into HBase and then to the data query by using remote sensing images of UAV as data source.Firstly,this thesis analyzes the application of cloud computing in GIS including the characteristics of two technologies.The advantages and technical difficulties after the combination of two technologies are presented.Secondly,this thesis introduces related theories and technologies including the principles of Hadoop architecture,tile data definition,the advantages of cloud computing technology and tile technology,and the concept and principle of tile Pyramid.The above introduction is to make technical preparation for the specific study and realization.Thirdly,based on the Hadoop architecture,this thesis designs the process of producing,storing and querying of tile data.The process is divided into the processing stage,the storage phase,and the query phase.In the processing stage,tile data are produced in parallel with the MapReduce parallel computing model.In this stage,the thesis proposes an improved task allocation method based on the cluster computing power and the tile data that needs to be generated.The method makes full use of cluster resources to improve the efficiency of tile formation.In the storage phase,the tile pyramid storage model is constructed in the HBase with Hilbert code.The process improves the efficiency of storage and index using the characteristics of the Hilbert curve.In the query stage,aiming at the segmentation characteristics of Hilbert space filled curves in query area,this thesis proposes an improved query algorithm based on multithreading,which improves query efficiency compared with traditional query methods.Finally,a Hadoop cluster is built as the experiment environment.In this cluster,a series of experiments is made,which proves that the methods proposed in this thesis have higher efficiency than other methods.