摘要
为了提高鬼成像的速度,提出了基于阈值分割的并行压缩差分鬼成像方法。首先通过设置合理的阈值筛选出涨落较为明显的数据进行采样;然后再通过并行分块降低图像维度,提高整体运算速度;最后通过压缩感知重构算法重构图像。通过对图"BUAA"的半物理仿真以及透射物体"光"的实验结果表明,与普通的二阶赝热光关联成像相比,该方法能以更少的采样次数和运行时间重构图像,可提升成像信噪比。
To improve the speed of ghost imaging, the parallel compressed differential ghost imaging method is pro- posed based on threshold segmentation （TS-PCGI）. Firstly, the large fluctuation of sampled data is selected by setting a reasonable threshold. Secondly, parallel computing divides the image into four parts. Finally, the image is restored through compressed sensing reconstruction algorithm. The results of a semi-physical simulated figure ＂BUAA＂ and ex- perimental object ＂light＂ show that this method can restore the object image with less sampling and time compared with conventional second-order correlated imaging. The signal noise ratio is improved.
出处
《光学技术》
CAS
CSCD
北大核心
2016年第4期289-293,共5页
Optical Technique
基金
总装预研基金(37708901)
关键词
并行计算
差分鬼成像
压缩感知
鬼成像
parallel computing
difference ghost imaging
compressed sensing
ghost imaging
