细胞筛选平台显微自动对焦系统研究

发布时间：2018-06-18 16:23

本文选题：细胞筛选平台 + 细胞图像处理　；参考：《河南科技大学》2017年硕士论文

【摘要】：国内对细胞筛选平台的研究较少,在生物细胞培养领域中一般采用96孔板作为培养基来培养细胞,由于96孔板内的细胞群落较小,肉眼难以分别,细胞筛选工作需要在显微镜下观测,面临着筛选检出率低、人工筛选易疲劳,手动对焦耗时耗力等问题。针对此问题,设计细胞筛选平台,研究细胞筛选平台显微自动对焦系统,包括细胞筛选平台的建立,运动控制系统的设计,显微自动对焦系统图像预处理以及自动对焦相关算法的优化。首先,在机构满足的前提下,细胞筛选平台通过控制单元的驱动,实现倒置生物显微镜载物台的自动进给和步进功能,其次,根据CCD所拍摄图像的清晰程度反馈至自动对焦处理单元,实现物镜的垂直进给,从而实现显微自动对焦系统的运行。论文分析细胞筛选平台显微自动对焦的工作原理和设计方法;研究显微自动对焦系统中对焦窗口的选取方法;传统的自动对焦区域选取方法对细胞显微图像进行对焦时,由于对焦区域的位置和大小固定,且细胞属于随机分布,并不总是处于图像的中心区域,导致对焦不准,忽略主体细胞群落,针对此问题,采用改进的鱼群算法,增加繁殖、淘汰行为,避免出现局部最优解的情况,提出一种基于鱼群算法的自动对焦窗口选取方法;通过对传统的对焦窗口选取方法和鱼群算法所获得的对焦窗口内图像,分别与改进后的鱼群取窗法进行对比分析,结果表明:改进后的鱼群取窗法所获得的图像包含更多的细节,且能自适应地寻找目标主体,并生成对焦区域;对焦区域内包含的图像经傅里叶变换函数处理后,其高频数量更多,得到了具有更高清晰度的图像。采用传统的自动对焦梯度函数评价算法在对显微图像进行自动对焦时,细胞显微图像中的细胞边缘灰度值梯度变化较小,对焦时易受到噪声影响。将Sobel梯度函数增加至4个方向算子模板;根据信号叠加原理,将符合Gaussian分布的Sobel4direction梯度函数在Brenner梯度函数上加权叠加,改变数据分布的离散程度,提出一种改进的Sobel梯度函数自动对焦评价算法,提高对焦精度;通过对传统自动对焦梯度函数和改进后梯度函数进行对比实验后,结果表明:改进后的自动对焦评价算法较传统的能更好的抑制噪声,电机在爬山算法的搜索对焦中具有更小的对焦搜索区间范围,获得的图像清晰度也更高。论文的研究结果对阐明细胞筛选平台的机理,揭示显微自动对焦系统中对焦精度的提高和对焦窗口的选取规律具有重要意义,可以为细胞筛选识别与分类奠定基础,在生物细胞工程方面具有重要的应用前景。
[Abstract]:In the field of biological cell culture, 96-well plate is generally used as the culture medium for cell culture. Because the cell community in the 96-well plate is small, it is difficult for naked eye to separate, because of the small cell community in the 96-well plate. The work of cell screening needs to be observed under microscope. It is faced with the problems of low detection rate, easy fatigue of manual screening, time and energy consumption of manual focusing, and so on. In order to solve this problem, the cell screening platform is designed, and the microscopic autofocus system of cell screening platform is studied, including the establishment of cell screening platform, the design of motion control system, Image preprocessing and optimization of autofocus correlation algorithm for micro-automatic focusing system. First of all, under the premise that the mechanism is satisfied, the cell screening platform realizes the automatic feed and step function of the inverted biological microscope platform through the drive of the control unit. Secondly, According to the clarity of the image taken by CCD, the automatic focus processing unit is fed back to realize the vertical feed of the objective lens, thus realizing the operation of the micro-automatic focusing system. This paper analyzes the working principle and design method of microscopical auto-focusing on cell screening platform, studies the selection method of focusing window in micro-automatic focusing system, and focuses on the cell microscopic image by traditional auto-focusing region selection method. Because of the fixed position and size of the focus region and the random distribution of the cells, they are not always in the center of the image, which leads to inaccurate focus and neglects the main cell community. In order to solve this problem, the improved fish swarm algorithm is adopted to increase the propagation. In order to avoid the occurrence of local optimal solution, an automatic focusing window selection method based on fish swarm algorithm is proposed, and the image in the focus window is obtained by traditional focusing window selection method and fish swarm algorithm. Compared with the improved fish group window extraction method, the results show that the image obtained by the improved fish group window extraction method contains more details, and it can find the target subject adaptively and generate the focus region. After the image contained in the focus region is processed by Fourier transform function, the high frequency quantity of the image is more, and the image with higher definition is obtained. Using the traditional automatic focusing gradient function evaluation algorithm, the grayscale gradient of the cell edge in the microscopic image changes little, and is easily affected by noise when focusing on the microscopic image. The Sobel gradient function is increased to four directional operator templates, and the Sobel4direction gradient function, which conforms to Gaussian distribution, is weighted on the Brenner gradient function according to the signal superposition principle, which changes the dispersion of the data distribution. An improved automatic focusing evaluation algorithm of Sobel gradient function is proposed to improve the focusing accuracy. The results show that the improved auto-focus evaluation algorithm can suppress noise better than the traditional one, and the motor has a smaller focus range and higher image clarity in the search focus of the mountain climbing algorithm. The results of this paper are of great significance to clarify the mechanism of cell screening platform, to reveal the improvement of focusing accuracy and the selection rule of focusing window in micro-automatic focusing system, and to lay a foundation for cell screening recognition and classification. It has important application prospect in biological cell engineering.
【学位授予单位】：河南科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：Q813;TP391.41

【参考文献】