遥感影像并行处理的数据划分及其路径优化算法

Parallel processing technologies have been widely applied to remote sensing images processing. While previous research has developed many parallel algorithms for processing images, few studies have been focused on synchronous parallel processing for multiple computing tasks when one copy of remote sensing image has many redundant backups under the cloud computing environment. To bridge the research gap, this research proposes a routing optimization algorithm for parallel processing of remote sensing image. Based on data segmentation, the method is developed to solve the dynamic routing optimization problem when applying the parallel technology to remote sensing image distributed storage and processing. Following the introduction of 8 definitions (e.g. model data state, model elements, relative information quantity and matrix mapping) and 6 properties (e.g. directed, transitive, reproductive, multi-dimensional properties), a mathematical model is proposed. Under the framework, the ratio of average computation costs is used as the flag to control horizontal or vertical parallel processing. In addition, typical examples such as quadtree index generation, and quadtree-based target detection are presented for illustrating the application of our model on parallel processing. Finally, through the experiments, we verify the effectiveness of the algorithm, discussing the characteristics and influential factors of the algorithm.