结合局部对比阈值与光晕移除的亮场细胞图像分割方法

Bright-field cell image segmentation method combining local contrast threshold and halo removal

针对亮场细胞图像存在对比度差、光晕破碎、细胞粘连,难以精确分割的问题,提出一种结合局部对比阈值与光晕移除的分割方法。首先,利用局部对比阈值检测并分割出含有细胞区域,再采用梯度迭代法进行光晕移除;然后,通过形态学操作与设置面积阈值进行优化处理,得到全局分割结果;最后,提取粘连细胞区域实现更精细的局部分割。实验表明:本文提出的方法能够移除细胞边缘的光晕伪影,解决粘连细胞的分割问题,实现亮场细胞图像的准确分割。

A method of bright-field cell image segmentation combining local contrast threshold and halo removal is proposed,aiming at the problems of poor contrast,broken halo and cell adhesion in bright-field cell images.Firstly,the region containing cells is detected and segmented by local contrast threshold,and then the halo is removed by gradient iteration method,then the global segmentation result is obtained by optimizing the morphological operation and setting the area threshold.Finally,the adherent cell region is extracted to achieve more precise local segmentation.Experimental results show that the proposed method can remove the halo artifacts of cell edge,solve the problem of segmentation of adhesion cells,and achieve accurate segmentation of bright-field cell images.