基于过完备字典稀疏表示的云分类研究

Cloud classification research based on over complete dictionary sparse representation

针对目前云类别自动识别方法较少的问题,提出了一种基于过完备字典稀疏表示的云分类的新方法.该方法用不同的云类型样本去建立自适应的过完备字典,提取字典特征,设计稀疏分类器,确定样本的云类型.仿真分析结果显示,本方法识别Ca,Cs&Cd,As&Ac,Ns&Cu,Cb云类型的准确率分别为100%,63.5%,90.3%,94.1%,98.2%,全局分类准确率为89.2%,优于支持向量机分类器和传统的稀疏表示分类器.

Aimed at the problem that automatic identification method for the cloud categories was less at present,a new method of cloud classification based on sparse representation of overcomplete dictionary was proposed. The method used different cloud types samples to establish an adaptive overcomplete dictionary,extracted dictionary features and designed sparse classifier to determine the type of cloud. The simulation analysis results showed that the classification accuracy of Ca,CsCd,AsAc,NsCu,Cb were 100%,63.5%,90. 3%,94. 1%,98. 2%,respectively. The overall classification accuracy was 89. 2%. The classification accuracy was higher than the support vector machine classifier and the traditional sparse representation classifier.