基于激光扫描共聚焦显微镜的相关图像处理技术研究

【作者】 李婧；

【导师】 郑筱祥； 张恒义；

【作者基本信息】 浙江大学 ， 生物医学工程， 2006， 硕士

【摘要】 激光扫描共聚焦显微镜具有成像分辨率高,能对细胞或组织内部进行非侵入式光学断层扫描而得到三维图像的优点,使其在生物医学邻域具有广泛的应用。细胞生物学研究对激光扫描共聚焦显微镜的应用提出了越来越高的要求,并且需要与之相适合的图像分析处理系统完成对共聚焦图像的处理、可视化及数据分析。 在实验室已经实现的针对激光扫描共聚焦显微镜的图像分析处理系统中包括:共聚焦数据场的预处理、基于Marching Cubes算法和交互体绘制算法的样本表面轮廓和内部结构的可视化以及一系列交互分析手段。在分析了共聚焦显微镜的原理和特点后,根据细胞生物学显微图像高分辨率的要求,为能得到更好的三维可视化结果,将自动聚焦和图像复原作为本文的主要研究方向。 针对激光扫描共聚焦显微镜扫描图像过程中自动聚焦能力的不足,提出了基于图像灰度和灰度梯度熵算子的二次聚焦评价方法,使共聚焦扫描显微镜在图像采集过程中实现自动对焦,提高图像的分辨率;同时针对共聚焦显微镜光路系统的特点及其对成像质量的影响,在对共聚焦系统点扩散函数进行理论推导后,采用基于统计学的最大似然估计方法对图像进行复原处理,用最大期望算法实现了该方法,改善了图像退化问题,有利于进一步的可视化和重建结果分析。 将以上研究成果结合到现有图像分析处理系统中,可以进一步提高图像三维可视化的质量以及结果数据分析的准确性,从而为更好地阐述其所揭示的生理学意义提供更为客观的依据。更多还原

【Abstract】 Laser Scanning Confocal Microscopy (LSCM) is now established as a valuable tool for obtaining high resolution images of a variety of Biological specimens. The development of Cellular Biology makes higher demand to LSCM, and requires appropriate image analysis and processing system to realize the process, visualization and data exploration.The image processing and analysis system aiming at the LSCM has been set up in our laboratory includes image preprocess, visualization of specimen contour and internal structure based on Marching cubes and interactive volume rendering as well as a series of interactive analysis technology. After analyzing the theory and characteristic of the LSCM, in order to obtain the better visualization result, we make auto-focusing and image restoration as the main study directions.For the disability of the auto-focusing of the LSCM, the study advances a fast image process algorithm to focus on an object image based on the image grey and entropy of the grey graduation. The auto-focusing technology applied in the LSCM will improve the resolution of scanning images. Analyzing the characteristic of the optical structure of the LSCM and it’s influence on the images quality, we carry the theoretic computation of the point spread function of the LSCM and adopt the expectation maximum arithmetic based on the maximum-likelihood estimation method to restore the blurred images. This step is very propitious to the visualization and result analysis.Combining above study results with the existed image analysis and processing system would further optimize the visualization quality and veracity of analysis of data, therefore provide more external reference to the physiological meaning expressed by the LSCM data.更多还原