基于改进NAS-RIF的三维显微图像盲复原  

The blind restoration of 3-dimensional microscopical images based on modified NAS-RIF

作　　者：杨运忠[1] 何小海[1] 陶青川[1] 

机构地区：[1]四川大学电子信息学院图像信息研究所,成都610064

出　　处：《光学技术》2008年第3期422-425,共4页Optical Technique

基　　金：国家自然科学基金资助项目(60372079)

摘　　要：对二维NAS-RIF算法进行了改进,并将它用于三维显微图像复原,分别用模拟样本和真实生物样本进行了实验。针对原算法对噪声有放大作用,提出了基于中值滤波去噪的改进。在支持域的确定、最小点的搜索上进行改进,提高复原质量。改进了原算法仅适用于均匀背景的不足,扩大了应用范围。实验结果表明,该算法在三维显微图像盲复原中具有较好的效果。The algorithm of NAS-RIF is modified and applied for 3-dimensional microscopical image restoration. Its performance is illustrated with synthetic and real biological data. To avoid the amplifying of noise in the original algorithm, an algo rithm of median filtering is introduced. To improve the effect of restoration, the algorithm of NAS-RIF is modified in the definition of support region and the research of the minimal point. Background rectification is developed for non-uniform background images to expand the applied scope of the original algorithm. Experiments show that the algorithm proposed has better result in blind restoration of 3-dimensional microscopical images.

关 键 词：图像盲复原 计算光学切片显微术(COSM) NAS-RIF 支持域 共轭梯度 中值滤波 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]