标注法在鼻咽双光子显微图像分割中的应用

APPLICATION OF LABELING METHOD TO THE SEGMENTATION OF TWO-PHOTON MICROSCOPIC IMAGE OF NASOPHARYNX

鼻咽细胞的双光子显微图像中含有丰富信息,借助计算机和图像处理算法进行分析处理,从而获取细胞的信息。首先以数学形态学方法对图像进行预处理,通过改进的函数在局部的每个对象形成单位极小值标记;接着在全局性的检测分割方法中,对单位极小值标记进行定位;最后对定位点的细胞过滤或填充还原其特性,这种算法能更快捷更准确地捕捉到图像中细胞的位置、个数等信息,获得较好结果。

For there is abundant information in a two-photon microscopic image of nasopharyngeal cells, the computers and image pattern processing algorithm could be adopted to analyze and process this sort of image so as to gain cells＇ information. In this paper, a mathematical morphology method is firstly introduced to conduct the image pre-processing, labeling each local object as minimum unit by an improved function. Secondly, we take up the position of minimum unit label by means of the global detecting segmentation method. Finally, the anchor cells could be filtered or filled in order to restore their characteristics. As it is faster and more accurately for this algorithm to capture the location, number of cells in the image and so on, it is able to get a better result.