Satellite image processing algorithm hybrid scheduling system and method

Abstract

The invention provides a satellite image processing algorithm hybrid scheduling system and method. The scheduling system comprises a management node, a management network, a Linux server, a Windows workstation, a data network and a network sharing storage device. The management node comprises workflow scheduling equipment which is used for creating, managing, scheduling and updating workflows; the management network is used for carrying out workflow job distribution and state collection between the workflow scheduling equipment and the Linux server and between the workflow scheduling equipment and the Windows workstation; the Linux server is used for executing a Linux algorithm; the Windows work station is used for executing a Windows algorithm; the data network is used for file transmission between the Linux server and the network sharing storage device and between the Windows workstation and the network sharing storage device; the network sharing storage device is used for storing files required by the Linux server and the Windows workstation for executing the algorithm. According to the invention, high performance is obtained, and automation of an image processing flow is realized.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a system and method for hybrid scheduling of satellite image processing algorithms. Background technique [0002] At present, the operating platforms of traditional satellite image processing algorithms are mostly PCs or workstations with Windows operating system. With the rapid increase in the number of satellites and the rapid increase in image resolution, image data has shown an exponential growth. The performance of traditional PC-based or workstation-based platforms is completely unable to cope with such a scenario. A cluster-based high-performance computing (HPC) system can connect multiple machines together through a high-speed network to aggregate computing power. Its performance is more traditional Several times, tens of times or even thousands of times of PC or workstation, and can be expanded almost linearly according to business growth, so more and more imag...

AI Technical Summary

Problems solved by technology

Another convenience is that there are already a large number of algorithms on the Windows platform. It takes a lot of money to port these algorithms to the Linux platform. Some algorithms need to be modified (such as some algorithms based on C / C++), and some algorithms need to be rewritten. (such as algorithms based on C#), some algorithms cannot be transplanted because they are too large

[0003] In order to use the algorithm under the Windows platform in the Linux cluster, the usual method is to manually copy the data to the Windows platform when necessary, and call the algorithm for processing, and then copy the data back to the Linux system after processing. The whole process is mainly manual Intervention, unable to automate the process

Images