摘要
提出宫颈癌细胞图象的数据采集,原始图象的平滑与边缘增强、分割与边界跟踪等处理方法。根据病理学家的经验,提取了细胞的21种特征,并基于相关性优选为5个特征。利用Fisher分类方法设计了分类器,实际判断结果良好。有关癌细胞图象识别的软件已研制成功。
Pressents a method for the data acquisition of cervical cancer cell image,the smooth, boundary enhancement, segmentation and edge track of original image.Based on the pathological experiments,21 features of the cell are extracted and optimized to 5.By Fisher method the classifier is designed. The software for cancer cell image has been programmed and used in practice successfully.
关键词
癌细胞
图象处理
计算机
模式识别
Cancer
Image processing
Data acquisition
Feature extraction
Feature selection
Pattern information processing systems ( PIPS )
