小麦染色体高分辨率G带的神经网络自动分析的研究  

AUTOMATIC ANALYSIS OF G BAND OF CULTIVATED WHEAT CHROMOSOME USING NEURAL NETWORK

作　　者：熊海涛[1] 胡匡祜[1] 孙艳[1] 李淑宇[1] 蔡念[1] 

机构地区：[1]中国科学院生物物理所,北京100101

出　　处：《生物物理学报》2000年第4期735-740,共6页Acta Biophysica Sinica

基　　金：国家自然科学基金资助项目(项目编号:39770226)

摘　　要：应用自组织特征映射神经网络原理与方法 ,建立了双层Kohonen神经网络 ,其第一层是从二维图象平面到二维特征平面的映射 ,用于染色体高分辨率带纹的提取和带纹参数的计算 ,第二层是高维特征参数阵列到二维聚类平面的映射 ,用于同源染色体的自动配对和分类。应用结果表明 ,该方法能快速、准确地实现对栽培小麦染色体高分辨率G带带纹图象的特征参数自动提取和自动配对。This paper presents a Double Kohonen Neural Network (DKNN) lased on the Self-Organize Feature Mapping theory to analyge chomosome automatiolly. The DKNN consists of two layers, each of which is a Kohonen network. The first layer maps a 2-d plane to a 2-d feature space to extract the high-resolution bands of chromosome and to compute the parameters of bands. The second layer maps the arrays of high dimension feature parameters to a 2-d plane to pair and classify the chromosomes automatically. The result indicates that this method can extract the feature parameters of G band in the chromosomes of cultivated wheat and then pair them automatically, rapidly and accurately.

关 键 词：自组织特征映射 神经网络 染色体高分辨G带带纹 染色体自动分析 

分 类 号：Q343.2[生物学—遗传学] S512.101[农业科学—作物学]