基于非降采样轮廓波和直方图变换的生物图像融合  

Biological Image Fusion Using a Nonsubsampled Contourlet Transform and Histogram Transform

作　　者：李添捷[1] 汪源源[1] 

机构地区：[1]复旦大学电子工程系,上海200433

出　　处：《中国生物医学工程学报》2010年第1期35-40,共6页Chinese Journal of Biomedical Engineering

基　　金：国家重点基础研究项目(2006CB705707);自然科学基金资助项目(10974035);上海市重点学科建设项目(B112)

摘　　要：随着分子成像系统的发展,图像融合在细胞生物学领域得到广泛应用,其中灰度结构图像与彩色功能图像的融合对蛋白质定位和功能的研究有重要意义。针对传统融合方法平衡结构信息和功能信息的难点,提出一种适用于生物图像融合的新方法。该方法基于非降采样轮廓波变换,得到源图像亮度分量的高低频分量,计算融合所需的权重矩阵,并由直方图变换对其做必要的调整,系数融合后经反变换得到融合的亮度分量,继而采用广义亮度-色调-饱和度法的快速实现算法,得到最终的融合结果。实验对117组来自约翰英纳中心的拟南芥细胞图像进行测试,采用视觉信息保真度,量化评估了荧光区和非荧光区中融合图像与源图像的相似程度。结果表明,新方法受荧光图像黑色背景影响较少,并保留了荧光图像中大量的细节信息,相比传统方法更为优越。The combination of the gray structural image and the color functional image is valuable for the localization and function studies of the protein. It is difficult to balance the structural and functional information injected to the fused image with conventional methods. Here we proposed a novel fusion method for uses in biological image analysis. After merging intensity components of original images based on the nonsubsampled contourlet transform and the histogram transform, the fast implementation of the generalized intensity-huesaturation （GIHS） method was adopted to obtain the fused image. The fusion effect is assessed for 117 sets of Arabidopsis images. Quantified indexes revealing the similarity between the fused product and original images were calculated inside and outside the fluorescent area. Results demonstrate the superiority of the proposed method to traditional methods.

关 键 词：图像融合 非降采样轮廓波变换 权重计算 生物图像 

分 类 号：TP391.7[自动化与计算机技术—计算机应用技术]