摘要
针对骨髓细胞图像的特点,采用数学形态学的方法对图像进行了处理,获得了不同类型细胞核的准确的边缘。对于获得的边缘图像,采用两级神经网络,利用基于神经网络的PCA算法获得图像的3个主分量,然后采用模拟退火算法和BP算法进行细胞的分类识别,获得了较好的识别效果。
Aiming at the characteristics of the marrow cell image, morphology methods to process image was adopted to detect accurate edges of different kinds of cells. Three principal components could be acquired by adopting two neural networks and using principal component analysis (PCA) based neural network. And marrow cells could be classed and recognized by using simulated annealing (SA) and BP network.
出处
《计算机工程与设计》
CSCD
北大核心
2006年第2期259-261,共3页
Computer Engineering and Design
基金
内蒙古高等学校科学研究基金项目(NJ04109)