鼻咽细胞的双光子显微图像中含有着丰富信息, 借助计算机和图像处理算法可进行分析处理。图象分割是双光子显微图象处理中的一项重要技术, 至今为止尚未形成一个最佳通用方法, 也没有定义出双光子显微图象分割的统一标准。本文首先采用噪声干扰法进行去噪, 采用低帽的变换等的数学形态学来增强鼻咽癌细胞图像, 使细胞更加容易分辨, 接着对几种经典边缘检测算法进行讨论比较, 紧接着根据鼻咽双光子显微图像的实际特征, 采取腐蚀算法求出鼻咽癌细胞边缘。然后进行区域生长定位细胞, 并采用一些改进的判别分析算法和区域面积算法对鼻咽癌细胞进行阈值分割, 获得较好结果。

There is abundant information in a two-photon microscopic image of nasopharyngeal cells, which can be analysed and processed by means of computers and image pattern processing algorithm. Image segmentation is an important technology and, at the same time, a hardship problem in the two-photon microscopic image segmentation processing. Up till now, there is no universal method or unified standard in defining the Two-photon microscopic image segmentation.Based on the noise denoising method, the writer of this paper adopted a mathematical morphology of Bottom-hat transformation to enhance the image of nasopharyngeal cancer cells, making the image easier to distinguish; Then, we tried to discuss and compare several classical edge detections, for example, Roberts edge algorithms, Prewitt edge algorithms, and Canny edge algorithms, etc. According to the actual characteristics of nasopharyngeal two-photon microscopic image, one could get the edge of nasopharyngeal cancer cells derived by means of corrosion algorithm. Furthermore, the article continued to probe a way to locate the cells in the regional growth, deviding the cells threshold segmentation with the help of improved discriminant analysis algorithm and regional area algorithm for a better effect.