基于随机采样的超高分辨率成像中快速压缩感知分析  

Fast compressed sensing analysis for super-resolution imaging using random sampling operator

作　　者：王中阳[1] 周燕飞[1,2] 张小伟[1,3] 沈灏[1] 李恩荣[4] 韩申生[4] 宓现强[1] 田立君[2] 彭玉峰[3] Wang Zhongyang Zhou Yanfei Zhang Xiaowei Shen Hao Li Enrong Han Shensheng Mi Xianqiang~ Tian Lijun Peng Yufeng(Shanghai Advanced Research Institute, Chinese Academy of Sciences, Shanghai 201210, China College of Electronics and Electronic Engineering, Henan Normal University, Xinxiang 453007, China Department of Physics, Shanghai University, Shanghai 200444, China Shanghai Institute of Optics and Fine Mechanics, Chinese Academy of Sciences, Shanghai 201800, China)

机构地区：[1]中国科学院上海高等研究院,上海201210 [2]上海大学物理系,上海200444 [3]河南师范大学电子与电气工程学院,河南新乡453007 [4]中国科学院上海光学精密机械研究所,上海201800

出　　处：《红外与激光工程》2017年第2期7-13,共7页Infrared and Laser Engineering

基　　金：国家重点研发计划(Y615852B01);上海市科委项目(Y431561D01,Y4340P1D02)

摘　　要：随着近年来超分辨成像技术的发展,基于单分子拟合的超分辨成像方法能够实现纳米尺度的空间分辨率,但这种方法的耗时较长,时间分辨率较差。成像重构时间较长主要受制于成像过程中每帧图像较低的荧光分子密度,所以需要足够多的采样帧数来重构一张图像。文中提出一种利用随机采样的快速压缩感知算法,结合分块压缩感知重构算法,最终能够在高分子密度的条件下获得较快的重构速度及较高的定位精度。Single molecule fitting-based super-resolution imaging methods achieve nanoscale image resolution but suffer from a long time resolution. The long acquisition time is limited by the low molecule density of fluorescent molecules that can be localized per imaging frame. In this paper, a fast compressed sensing method was proposed to use random sampling operator, together with the block compressed sensing reconstruction method, have a good performance of localization accuracy under high molecular density and reconstruction speed.

关 键 词：压缩感知 超分辨率成像 随机采样 

分 类 号：TP731[自动化与计算机技术—检测技术与自动化装置]