基于图割算法的宫颈细胞分层次分割

HIERARCHICAL SEGMENTATION OF CERVICAL CELLS BASED ON GRAPH CUT ALGORITHM

宫颈细胞分割在宫颈细胞形态学研究中是十分重要且具有挑战性的环节。提出一种基于图割的分层次高效分割宫颈细胞方法。对获取宫颈细胞图像进行灰度变化处理求得灰度直方图,根据直方图结合OTSU求解最佳分割阈值,根据最佳分割阈值对图像进行初次分割去除背景,得到单个细胞目标;而对粘连、重叠、较为复杂的细胞图像再利用图割算法进行二次分割获取单个目标细胞。实验结果表明,该算法不仅有较好的分割效果而且效率也很高。

Cervical cell division is an important and challenging part in the study of cervical cell morphology. This paper put forward a hierarchical efficient segmentation method for cervical cells based on graph cut. The gray scale histogram was obtained by processing the gray change of the cervical cell image. The optimal segmentation threshold was solved according to the histogram and OTSU. The image was firstly segmented according to the optimal segmentation threshold to remove the background, and a single cell target was obtained. The adhesive, overlapping and complex cell images were segmented twice by graph cut to obtain a single target cell. Experimental results show that the algorithm has good segmentation effect and high efficiency.