摘要
各种类型云的辐射特性以及分布情况,对大气收支平衡以及天气气候都有重大影响,对云进行正确分类是遥感领域的重要应用和研究热点。文章基于对卫星云图进行自动准确识别和分类研究的前提,通过介绍几种特征提取和选择方法,以及介绍无监督、有监督和神经网络3类云分类研究常用分类方法,对国内外近几十年来所做的卫星云图分类研究进行综述介绍。并简要介绍了云分类结果的评价方法,对分类研究的结果进行定性讨论。
Radiation features and the distribution of kinds of clouds have significant influences on atmospheric radiation balance and climate.Therefore,how to classify clouds in satellite images is an important application as well as study focus in remote sensing field.The article introduces not only some methods of feature selection and extraction,but also unsupervised,supervised and neural network classifiers,which always used in cloud classification studies.Then it reviews studies of cloud classification over the past decades.Evaluation methods are discussed briefly with classification results in some studies.
出处
《电子设计工程》
2011年第10期189-192,共4页
Electronic Design Engineering
基金
国家自然科学基金(41005024)
山东省博士基金项目(BS2010DX034)
关键词
云分类
特征提取和选择
模式识别
神经网络
cloud classification
feature extraction and selection
neural network
pattern recognition