The traditional centralized single mode becomes a “bottleneck” of remote sensing image processing which cannot meet the needs of future remote sensing image processing development. Fortunately, the distributed parallel computing has provided a turning point to the quick calculation of remote sensing image. This paper presents the cluster computing environment based on the MPI, and advances a project of a parallelized design to the gray level co-occurrence matrix algorithm. Moreover, the experimental data, which is due to the parallelized algorithm running in the cluster, is recorded and analyzed in several respects such as different nodes, time, speedup, efficiency and so on. The analyzed result shows that parallel computing cluster based on MPICH can efficiently improve the speed of remote sensing image processing in the case of more complex algorithms. However, when the number of node increases, the consuming time decreases, and the efficiency will decrease at the same time. So, it is rather important to keep the balance between performance and efficiency. The nodes can not be continuously added into computing, when the consuming time can be accepted.
An improved parallel hybrid bi-conjugate gradient method suitable for distributed parallel computing
