高斯型点扩展函数估计的最近邻算法  

Nearest-neighbors subtraction algorithm based on Gaussian point spread function estimation

作　　者：李蕊[1] 陶青川[1] 何小海[1] 罗代升[1] 吕成淮[1] 

机构地区：[1]四川大学电子信息学院,四川成都610064

出　　处：《光电工程》2007年第6期97-101,111,共6页Opto-Electronic Engineering

基　　金：国家自然科学基金资助项目(60372079)

摘　　要：本文针对计算光学切片中的最近邻算法提出了一种改进算法。通过小波变换计算出高斯点扩展函数的方差值,再根据相邻图像成像及高斯函数特性,得出所需的高斯型层间点扩展函数。同时,文章还给出了两种高斯型层间点扩展函数方差的获得方式及获得过程,对最近邻算法中的加权因子的取值范围做出了讨论,对传统的最近邻算法做出了改进。实验表明,本算法能够更有效地复原符合最近邻要求的切片图像。在点扩展函数未知的情况下,复原效果要优于传统方法。An improved nearest neighbor subtraction algorithm was presented and applied in the Computational Optical Section Microscopy （COSM）. Firstly, the variances of Gaussian Point Spread Function （PSF） were calculated by using wavelet transform. Secondly, according to the characteristics of adjacent slices and the Gaussian function, the Gaussian PSF between layers was obtained. Thirdly, two estimation methods and processes acquiring the variances of the Gaussian PSF between layers were presented. Finally, the range of possible value of weighted factor was discussed in details. Experiments show that compared with the traditional nearest-neighbor subtraction algorithm, the new algorithm can restore the blur section images which are in accordance with the nearest-neighbors subtraction algorithm＇s standard more availably. The restoration results of presented algorithm are better than that of traditional algorithm when the whole or part of PSF is unknown.

关 键 词：计算光学切片 最近邻算法 高斯型点扩展函数 小波变换 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]