基于静态小波变换的变透明度法融合GFP荧光与相衬图像

绿色荧光蛋白荧光图像与相衬图像的融合对蛋白质功能的研究和亚细胞结构的定位有重要价值。针对成功用于遥感图像融合的ARSIS概念下多尺度融合算法融合荧光图像与相衬图像时易产生伪像的缺点，提出变透明度的概念，根据直观视觉效果设计一组函数，为融合图像的每一像素分配尺度透明度，并基于静态小波变换的分解结果对源图像进行融合。先通过30组图像的融合实验，估计出所设计函数的相应参数，然后用于对另外117组图像的融合测试，分别计算融合结果的荧光区和非荧光区与荧光图像、相衬图像的质量指数Q和高频相关系数HPCC这两个表征融合效果的特征参数。结果表明，相比常用的融合算法--透明度法、棋盘格法和静态小波替换法，本方法在保持交互可变透明度的同时，提高了融合结果细节的清晰程度，降低了荧光图像背景对融合结果的影响。

Fusion of GFP fluorescent and phase contrast images using stationary wavelet transform based variable-transparency method

It is valuable to merge the green fluorescent protein (GFP) fluorescent image and its corresponding phase contrast image for the localization of the protein and sub-cellular structures. When ARSIS concept based methods which have been successfully applied to merge remote sensing images are used for the fusion of GFP fluorescent and phase contrast images, there may exist many artifacts. To overcome this limit, a variable-transparency concept is proposed to design a set of functions based on visual effects. After each pixel is assigned the multi-scaled transparency, original images can be synthesized from their results of stationary wavelet transform (SWT). Firstly 30 groups of images are used to estimate parameters involved in the function set and then 117 groups are assessed by means of qualitative tests. Quantified indexes which are related with the fused product and original images, such as the quality index Q and the high frequency correlation coefficient HPCC, are respectively calculated for inside and outside of the fluorescent area. Results demonstrate that compared with common algorithms, including the transparency method, the chessboard method and the SWT substitution method, this proposed algorithm can keep the interactive user-transparency and improve the definition of the fused product while reduce the interruption from the dark background of fluorescent images.