荧光3维解卷积成像及其正则化方法  

3-D deconvolution for fluorescence imaging and its regularization method

作　　者：叶大田[1,2] 方滨[1,2] 钱翔[1] 

机构地区：[1]清华大学深圳研究生院,深圳518055 [2]清华大学生物医学工程系,北京100084

出　　处：《清华大学学报（自然科学版）》2014年第11期1391-1395,共5页Journal of Tsinghua University(Science and Technology)

基　　金：国家自然科学基金项目(81201165)

摘　　要：解卷积荧光成像方法由于能够提高图像质量,因此在生物领域得到越来越广泛的应用。该文基于RichardsonLucy(R-L)算法,提出全变分(TV)正则化的盲解卷积算法。采用根据实际实验系统参数仿真的球模型以及真实的荧光图像,来比较正则化前后图像的复原效果。结果表明:R-L算法可以去除由于点扩散函数引起的大部分模糊,而随着迭代次数的增加,TV正则化算法能抑制噪声的放大。TV正则化的盲解卷积算法在复原实际的生物图片中效果较好。The deconvolution fluorescent imaging method has become more and more popular in biological applications due to its improved image quality. This paper presents a Richardson-Lucy （R-L） algorithm with blind deeonvolution with total variation （TV） regularization. The simulated sphere model is used with real experimental parameters and real fluorescent images to compare the restoration results before and after regularization. The results indicate that the R-L algorithm can remove most of the blurring caused by the point spread function while the TV regularized methods avoid the noise amplification with increasing iterations. The TV regularized blind deconvolution does particularly well in restoring real biological images.

关 键 词：解卷积 荧光成像 正则化 图像复原 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]