层次聚类在细胞图像分析中的应用

APPLICATION OF HIERARCHICAL CLUSTERING IN CELL IMAGE ANALYSIS

传统的细胞图像分析是从细胞图像分割或者边缘检测的角度进行的,这些方法是通过对细胞图像像素变化的分析达到对细胞图像进行分析的目的;而层次聚类方法是从对细胞图像像素进行聚类分层的角度达到提取细胞的目的。细胞图像像素的聚类是结合了半边图的原理,将每个像素或者聚类后的一类像素看作是半边图中的顶点,细胞图像像素之间的距离、像素值差等看作是半边图的半边,将符合条件的半边结合成边,能连接成边的顶点归为一类。通过对细胞图像像素的层层聚类最终得到完整的细胞图像。

Traditional cell image analyses are carried out from the perspective of cell image segmentation or edge detection,these methods reach the target of analysing the cell image by analyzing the variation of cell image pixels;while the hierarchical clustering method reaches the goal of extracting cells by clustered stratification of the cell image pixels.In this paper,the clustering of cell image pixels combines the theory of half-edge image,in which every pixel or a class of pixel being clustered is regarded as the vertex of the half edge image,the distance between cell image pixels and the difference of pixels are regarded as the half edge of the half edge image.Those half edges meeting the condition will be connected to line to an edge,and those vertexes eligible to line to edge will be grouped to a class.By layer upon layer clustering the cell image pixels,a complete cell image can be eventually obtained.