基于距离变换的粘连细胞分割方法  

Separate Algorithm for Overlapping Cell Images Based on Distance Transformation

作　　者：游迎荣[1] 范影乐[1] 庞全[1] 

机构地区：[1]杭州电子工业学院自动化分院,杭州310018

出　　处：《计算机工程与应用》2005年第20期206-208,共3页Computer Engineering and Applications

基　　金：国家自然科学基金(编号:60374047);浙江省自然科学基金重点项目:光化学复合敏感模式识别与集成处理(编号:ZD0205)资助研究

摘　　要：在医学细胞图像中,经常会有细胞重叠和粘连的现象。将粘连在一起的细胞群单个分开是医学细胞处理中一项重要而困难的问题。对这一情况,文章提出了一种有效的分割算法。先用最大类间方差法对图像进行二值化,然后对图像作距离变换,将像素点位置信息转化为灰度信息,再利用分水岭算法,找到粘连边界点,从而将细胞群分开。实验证明,该算法能成功地分离粘连细胞。The overlapping phenomenon often appears in the medical cell images.Separating the cell cluster into single ones is great important and difficult in medical cell image processing.In allusion to this phenomenon,an effective segementing algorithm is presented to solve this problem.Firstly,threshold image by means of OTSU,then distance transform the threshold image for transform the position information to the gray information,whereafter,mark the edge points by using watershed algorithm and then separate the cell cluster.The experiment result shows that this algorithm can separate the cell cluster successfully.

关 键 词：最大类间方差法 距离变换 分水岭算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]