基于小波变换与纹理特征分析的地基云图识别  

Classification of ground-based cloud image based on wavelet transform and texture analysis

作　　者：陈锦源[1] 高太长[1] 刘磊[1] 韩文宇[1] 

机构地区：[1]中国人民解放军理工大学,南京211101

出　　处：《气象水文海洋仪器》2014年第1期1-4,共4页Meteorological,Hydrological and Marine Instruments

基　　金：国家自然科学基金(41205125):复杂天空云状的特征描述及自动识别研究

摘　　要：云状是云观测的三大要素之一,正确识别云状有助于分析大气运动状态,为天气预报提供重要依据。文章采用了图像处理的方法对全天空红外云图进行初步云状识别的研究,利用小波分析与纹理特征(灰度共生矩阵)分析相结合的方法提取所需识别云类型的特征参数,设计并建立适合云状识别的分类器。所做分析和研究工作表明,对比单一纹理特征分析和利用小波分析与纹理特征分析相结合的方法对所选全天空红外云图的识别结果,后者具有较高的准确率。Cloud type is one of the three elements of cloud observations. And correct identification of cloud type will have a help for analyzing the state of atmospheric motion,providing an important basis for the weather forecast. This paper introduces a preliminary research on cloud classification for whole- sky infrared images by means of the method of image processing. The required characteristic parameters for cloud classification are extracted based on the wavelet transform and texture analysis （GLCM）. Then, the suitable classifier for cloud classification is designed and built. The analysis and research work show that comparing a single texture analysis method with the wavelet transform and texture analysis method for the recognition of selected all-sky infrared cloud, the latter has a higher accuracy rate.

关 键 词：全天空红外云图 小波分析 纹理特征 

分 类 号：P412.15[天文地球—大气科学及气象学]