摘要
详细介绍了一种基于显微多光谱宫颈图像的迭代分割方法,该方法将多光谱分析应用到宫颈显微图像分割中。由于显微宫颈细胞图像十分复杂,普通的分割方法不能解决分割问题,该文给出了一个较好的解决方案。算法首先对大量多光谱波段进行筛选,接着利用细胞的多光谱信息并采用迭代分割算法不断优化分割精度,经过大量实验证明算法具有分割速度快、精度高、算法稳定性好等特点。文中的多光谱波段的筛选对于利用多光谱手段进行其他图像分割都有很大的借鉴意义。
This article presents a new iterative segmentation method for multispectral micrographs of cervical cells,and the method absorbs the advantages of multispectral analysis.Because of the complexity of cervical cells micrographs,common segmentation means are generally invalid,where the means discussed in this article gains substantial progresses.Firstly,A selection was made in a large amount of spectral bands,and then based on multispectral information,segmentation runs iteratively to optimize the results.Confirmed by abundant segmentation experiments,this means is fast,accurate and steady.Meanwhile the process of spectral band selection is also valuable for segmentations of other multispectral graphs.
出处
《计算机工程与应用》
CSCD
北大核心
2005年第10期191-193,206,共4页
Computer Engineering and Applications
基金
国家科技攻关"十五"项目:光谱成象分析系统的研制与开发(编号:2001BA210A02)
关键词
图像分割
多光谱
宫颈细胞
迭代分割
image segmentation,multispectrum,cervical cell,iterative segmentation
