1. 湖南大学计算机与通信学院,湖南,长沙,410082
2. 中南大学信息科学与工程学院,湖南,长沙,410083
3. 清华大学软件学院,北京,100084
4. 湖南大学计算机与通信学院湖南长沙,410082
5. 中南大学信息科学与工程学院湖南长沙,410083
6. 清华大学软件学院北京,100084
纸质出版:2005
移动端阅览
刘相滨, 邹北骥, 孙家广. 菌群细胞图像分离算法研究[J]. 电子学报, 2005,33(6):1056-1059.
LIU Xiang-bin, ZOU Bei-ji, SUN Jia-guang. A New Algorithm for Separating Cell in Bacteria Image[J]. Acta Electronica Sinica, 2005, 33(6): 1056-1059.
颗粒图像分析中聚堆目标的分离对目标的计数及特征的提取非常重要.现有分离算法都要求聚堆目标粘连处凹陷比较明显
并且/或者存在灰度局部最小边缘.菌群聚堆细胞大小不一、聚堆形态各异
多数聚堆细胞并不具备上述条件.文中提出了一种基于聚堆区域轮廓跟踪、删除的分离算法
依据跟踪"虫"在跟踪过程中遇到已跟踪过轮廓点的情况判断分离的进行.实验结果表明算法对菌群聚堆细胞分离的成功率要高于现有算法10%以上.
Separating objects touching and overlapping with each other into single ones is very important for the counting and feature extracting.Most of the existing separating algorithms are based on two hypotheses:(1)the touching objects tend to form an acute angle on the contour where objects touched
and/or (2)there would always be a lower optical density at touching points.For colony image
the bacteria cells vary in sizes and touching manners.Many touching cells are short of above hypotheses.A new separating algorithm based on 8-connected contour tracking is proposed in the paper.The boundary of touching-cell was tracked and deleted layer by layer
and judgment was made by the tracking result to determine whether the touching-cell should be separated.The experiment results show the success ratio to separate bacteria touching-cell is 10% higher than the existing algorithms.
0
浏览量
1257
下载量
3
CSCD
关联资源
相关文章
相关作者
相关机构
京公网安备11010802024621