基于优势背景色的细胞图像综合分割方法

摘要

The segmentation method for breast cell color image based on mathematical morphology is presented. On the basis ofthe histogram distribution of cell color image RGB, we selected the RGB channel with the largest pixel percentage in the lighted area, which presented images with obvious contrast and smooth background. Then, we conducted a morphological reconstruction of shaded area for the RGB channel and conducted background filtering with the foreground shape intact. We performed histogram matching and further narrowed down background gray level and finally we carried out dynamic threshold segmentation. The experiment revealed that the proposed method could be carried out automatically with excellent effect of noise eliminating and cell form maintained.%提出了基于优势背景色的数学形态学综合分割算法对彩色细胞图像进行分割.根据彩色细胞图像RGB这3个通道的直方图分布,选择亮区像素比例最大的一个通道,该通道下图像反差较大,背景较平坦.对该通道进行灰度形态学重构,在保留前景形状的同时,对背景进行滤波,再进行直方图规划操作,进一步压缩背景灰度级,最后进行动态阈值分割.实验结果表明,该方法可自动进行,去除噪声效果好,分割出的图像很好地保留了细胞形态.