节点文献
基于图像区域特征的细胞识别方法及实现
A Cell Recognition Approach Based on Image Area-Character and Its Realization
【摘要】 先提取目标物体的面积、中心矩、圆形度、细长度、核浆比等区域特征。然后采用单原型模式表征方法,利用集群分析技术中最小距离分类器,构造判别函数,并依此作为识别类型的依据。最后,通过对鳞细胞的区域特征进行计算比较,选择出能够较明显地辨别出类别的特征向量,建立判别函数并对结果的可信度进行了讨论。
【Abstract】 Firstly,this paper picks up area characters of the object including area,bary centeric moment,the round degree,slightness degree and nucleoplasm ratio of the geometrical modality.The signmethod of single antetype and the least space sortimplement are used to constract a distinguish funtion.Finally,by comparing and computing area character of Squamouscell,the author elects characteristic vectors through which sorts squamous cells can be discerned obviously,establishes the distinguish function and discusses the reliability of results.
【基金】 江苏省高校科学研究项目(02KJD520005)
- 【文献出处】 南京气象学院学报 ,Journal of Nanjing Institute of Meteorology , 编辑部邮箱 ,2003年05期
- 【分类号】R310
- 【被引频次】24
- 【下载频次】410