Design and implementation of cluster-based platform for Remote Sensing Information Computation

Remote Sensing Information Computation has features of large amount of image data and complex algorithms, many applications require real-time response, but mainstream computers are often unable to meet their performance requirements. Regarding to this problem, a cluster-based computing platform for remote sensing information computation was designed and implemented through interconnecting the existing computing devices by network, and in support of MPI and other communication software, cluster management module which enables node management, task management, and real-time monitoring and other functions was developed. On this basis, the parallel algorithms of re-projection, mean-shift multi-scale segmentation were achieved; the efficiency and performance of the cluster were tested. Experiment results show that the system can significantly improve the efficiency of remote sensing information computation with a low price.