The Research And Application Of Remote Sensing Algorithm Driven Strategy In Remote Sensing Satellite Product Production System

While taking part in the remote sensing algorithm driven module of product production subsystem which was one part of the national high resolution major special” Earth surface system scientific research application demonstration system”,the author found that with the increase of remote sensing algorithm complexity,the internal dependence of the remote sensing algorithm was not clear,the efficiency of algorithms driving was low and the ability of large scale processing was weak.The algorithm generation tree could only describe one-to-one or one-to-many algorithm dependencies relationship at present,and then appeared low utilization rate of CPU when driving the remote sensing algorithm to produce remote sensing products.In order to solve these questions,this paper used the graph theory,Dijkstra,the large-scale and parallel processing technology of remote sensing data and other related technology,focused on the research of description of internal dependencies about complex remote sensing algorithm,algorithms driven efficiency and large scale remote sensing products in cluster environment.(1)Description of complex remote sensing algorithm dependence relationship and the design of the algorithms driven model.Come up with using the weighted directed graph to describe the internal dependence relationship of remote sensing algorithm,made the inner dependence relationship of algorithm much clearer,laid foundation for visualization of the inner dependence relationship and execution progress of algorithm.The weight of each sedge was set to the number of the algorithms those were dependent on this side and the weight of each sedge showed the dependency degree of the fixed-point algorithm.,at the same time it could also be used as a parameter to measure the complexity of a task.Then used Dijkstra to traverse the algorithm graph could fond the algorithm node that needed to execute priority.The algorithms those are no dependencies could execute in parallel.The simulation results showed that this method can improve driving efficiency of algorithms and the utilization rate of CPU.(3)The algorithms driven model was applied in the production subsystem.In order to verify the practical effect in country major projects,the algorithms driven model was applied in the production subsystem of national major projects.Realized adaptive parallel processing mechanism by building public computing space and using computational storage integration technology.At last,adopted the cluster technology to improve the large scale processing capacity of system,and supported multi-user and multitasking.Eventually showed that the driving strategy could improve the efficiency of the production algorithm of remote sensing satellite effectively.