基于多核CPU的遥感去雾并行算法研究

Study on Parallel Algorithm Method Based on Remote Sensing Defogging of Multi-core CPU

针对数据量较大的遥感影像进行去雾处理时出现的运行效率低下或者是计算无法进行的问题,提出了多核CPU并行去雾算法。首先对影像进行分块分割,再根据滤波影响域对分块影像进行扩边,并建立影像索引,然后将分块影像分配给不同的CPU内核进行去雾处理,最后利用影像索引将去雾处理后的分块影像进行合并。实验表明,经本文改进的并行去雾算法与传统暗通道去雾算法的去雾效果一致,而算法的执行效率得到了提高,具有很强的实用价值。

Aiming at solving these problems of low operation efficiency or no-performed calculation when remote sensing image defogging with a large amount of data is conducted, the parallel defogging algorithm method of multicore CPU is put forward. Firstly, the image is divided into small blocks which are expanded edges according to the influenced domain of filtering and builds the image index. Then these block images are assigned to different CPU cores to conduct the processing of defogging. Finally, these block images after the processing of defogging are merged by using image index. Experimental results show that the parallel defogging algorithm method improved by this paper has the same defogging effect with the traditional defogging algorithm method of dark channel, but its efficiency has been improved and has strong practical value.