Study Of Remote Sensing Image Processing Methods Based On Cloud Computing Platform

With the highly engagement in the field of earth observation technology of our country,remote sensing is developing rapidly toward high spatial,high spectral and high temporal resolution and the applications of different kinds of remote sensing satellites are constantly expanding.The high resolution remote sensing images lead to the huge increasement of remote sensing data.Therefore,how to manage and calculate these massive remote sensing data,and then to provide efficient remote sensing image processing services on this basis,both have become a generally recognized challenge in remote sensing.In the meantime,the development of information technology especially the high-speed rise of cloud computing in recent years is providing a new feasible solution for remote sensing image processing in large quantities.First,this paper focused on remote sensing image processing technology and core technologies of cloud computing.And the conventional process and algorithms of remote sensing image processing were organized.Then an MPI remote sensing image parallel processing model based on GDAL was adopted to solve the problem of inefficiency in massive remote sensing image processing.Typical algorithms including radiometric calibration,atmospheric correction and feature extraction were designed and applied with this model.Then the above functions of remote sensing preprocessing and processing were encapsulated.Second,data resource management technology,parallel data processing technology,network communication and virtualization technology and Web services building technology were summarized to build a proper structure of cloud computing platform.Based on these technologies and according to the characteristics of the remote sensing application itself,the overall framework of cloud computing platform was designed,and specific technologies of Network File System,MPI multi-process parallel processing model,SSH data communication and transmission and Web services visualization based on PHP framework were adopted to deploy the remote sensing cloud computing platform.Thirdly,functions of remote sensing parallel processing were integrated in the platform.Finally,the RS cloud computing platform was successfully configuredand on a specific rack-mounted server.From the platform a SaaS remote sensing image processing web application was provided for clients.Experimental measurement for accuracy and performance of remote sensing services was run on the platform.The results showed that the remote sensing image parallel processing functions were accurate and the acceleration rate was as as high as 313%compared with a PC environment.The effects on processing efficiency with different resource utilization of the platform were analyzed.Thus a reference model was provided for cloud computing used in remote sensing image processing applications.This research provides a common and effective solution for remote sensing image processing service based on a cloud platform and has value in massive remote sensing image processing.