国产卫星遥感影像多核并行自适应正射纠正方法

<正>射纠正是生成正射影像的关键步骤。随着国产卫星不断发射升空,国产卫星遥感影像的数据量急剧增加,正射纠正的处理速度越来越受到关注。传统的卫星遥感影像正射纠正多采用串行固定分块策略进行纠正,效率较低,单景处理时间长,单位时间内正射纠正的生产达不到要求,而GPU高性能加速对计算机硬件的要求较高,在实际生产中并没有很好地使用。本文在分析正射纠正算法并行性的基础上,提出了多核并行分块自适应正射纠正方法,能够在台式机和普通笔记本上高效计算,极大地提高了正射纠正的效率,获得了一定的加速比,很好地满足了实际生产要求。

Multi-core Parallel Self-adaptive Ortho-rectification Method for Domestic Satellite Remote Sensing Images

Ortho-rectification is a key step in the process of generating orthophoto map.With the constantly launch of domestic satellites,there is a sharp rise in the amount of domestic satellite remote sensing image data,so the processing speed of ortho-rectification is getting more and more attention.Traditional satellite remote sensing image ortho-rectification adopts the serial fixed partition strategy.However,this method is not in high efficiency and the processing for single image is time-consuming.It is hard to satisfy the ortho-rectification production requirements in unit time.Due to the high requirement for computer hardware,the GPU high-performance acceleration actually is not in good use in practical production.In this paper,based on the parallelism analysis of ortho-rectification algorithm,multi-core parallel block self-adaptive ortho-rectification method is proposed.The proposed method is highly efficient in conducting computations in desktop and ordinary laptop,which greatly improves the efficiency of ortho-rectification and can obtain a certain degree of speedup ratio.The proposed method satisfies the practical production requirements well.