摘要
为了提高光电子鼻的气体区分能力,对7种气体样本进行了检测和分类实验。将敏感阵列的响应图像从RGB颜色空间向其他6种典型的颜色空间进行了转换,并通过增L减R(L-R)搜索算法,在欧式距离样本可分性判据的基础上,从新生成的7种颜色空间中优选了18个颜色通道组成融合空间。主成分分析和欧式距离可分性判据对比表明,电子鼻在融合颜色空间的气体响应特征向量的类别可分性要显著优于其他颜色空间。
Aimed at improving the gas discrimination capability of an optical electronic nose,seven kinds of gases are tested and classified.The response images of the sensitive arrays measured in RGB color space are transformed to other six typical color spaces.From the new seven color spaces,a L-R search algorithm based on an euclidean separability criterion is performed to pick out 18 optimal color channels to make a fusional color spaces.A joint analysis of principle component analysis(PCA)and the euclidean separability criterion comparison shows that the feature vectors upon the seven kinds of gases are much discriminable in the fusional color space than in other color spaces.
作者
唐忠林
TANG Zhonglin(School of Aeronautical Engineering,Shaanxi Polytechnic Institute,Xianyang,Shaanxi 712000,China)
出处
《自动化应用》
2024年第11期38-41,共4页
Automation Application
基金
陕西工业职业技术学院2023年科技创新专项项目(2023YKZX-001)。
关键词
颜色
特征优化
空间融合
搜索算法
电子鼻
color
feature optimization
space fusion
search algorithm
electronic nose
