GMS卫星红外云图强对流云团的识别与追踪  

RECOGNITION AND TRACING OF SEVERE CONVECTIVE CLOUD FROM IR IMAGES OF GMS

作　　者：白洁[1] 王洪庆[1] 陶祖钰[1] 

机构地区：[1]北京大学地球物理系暴雨监测与预测国家重点实验室

出　　处：《热带气象学报》1997年第2期158-167,共10页Journal of Tropical Meteorology

基　　金：国家自然科学基金

摘　　要：利用ＧＭＳ静止气象卫星探测所得到的红外云图资料，用计算机图象学技术来研究强对流云团的识别和追踪，以求能够准确、及时地发现并追踪象强对流云团这种能够造成强烈灾害的天气系统。对红外云图的图象处理，提出区域平滑滤波和阈值剔除相结合的强对流云团过滤算法，对于过滤出的强对流云团，应用图象处理中轮廓编码法的Ｔ算法和ＩＰ算法提取出它的边界、边界初始点。强对流云团的描述包括其重心、特征面积、矩特征量以及Ｒ形状描绘子等特征量。由描述强对流云团的特征量，应用模式识别和模式匹配技术对强对流云团进行追踪。给出粗对比分析和细对比分析两个模式匹配过程，粗对比分析包括搜索区设定和面积识别两步，而细对比分析由Ｒ形状描绘子匹配。It is very important to find and trace the weather system,such as severe convective clouds(SCC),which cause severe calamity.This paper describes the finding and tracing of SCC using computer image technique for GMS IR image. In this paper we put forward a segment smoothing filter algorithm and threshold algorithm to filter SCC in the IR image preprocessing. With respect to the filtered SCC, the T algorithm and IP algorithm of outline coding method in image processing are used to extract the boundary line and the first point in the line from SCC. The description of SCC includes four characteristic quantities, i.e. center of gravity, cloud size, moment invariant M and shape descriptor R. In this paper pattern recognition and pattern match technique are used to trace the SCC. Two pattern processes of recognition and matching, i.e. rough and fine comparative analysis,are provided . Rough comparative analysis includes the establishment of search area and size recognition, and fine comparative analysis is composed of shape descriptor R match, moment invariant M match and correlation brightness analysis.

关 键 词：模式识别 红外云图 强对流云图 云团 气象卫星 

分 类 号：P455[天文地球—大气科学及气象学] P412.15