欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
提出了一种重叠细胞检测的新方法,该方法针对细胞图像处理中因细胞重叠而引起的过分割与欠分割现象这一难点,引入基于Mumford-Shah模型的水平集方法得到细胞的初始轮廓,采用EP（erosion-propagation）和PL主曲线相结合的图像分析方法得到目标的精确轮廓线。使用基于目标形状的图像配准方法和基于目标轮廓的分段特征分类器完成重叠细胞的检测。应用该方法对600张随机抽取的细胞图像中9374个细胞进行检测,实验正确率为91.83%,结果表明该方法能够有效地判断细胞是否重叠和重叠类型,为后续分割工作提供依据。
英文摘要：
Propose a new method for overlapping cell image detection.In order to solve problem caused by the over-and under-segmentation,a level set method based on Mumford-Shah model is applied to gain the initial contours and the accurate contours are provided by unifying the erosion-propagation（EP） algorithm and the PL principal curves.After that,registration for shape models and classification based on the segmented contour features are applied for the detection of overlapping cells.Experimental results on 9374 cells in 600 randomly selected images with a correct rate of 91.83% indicate that the overlapping cells are well detected and classified.The work shows great assistance for the image segmentation afterward.
同期刊论文项目
同项目期刊论文
期刊信息
位置：