基于卫星云图的台风云系特征提取算法研究

Segmentation of Tropical Cyclone Cloud Systems with Compositive Algorithms

利用静止卫星观测数据,进行了滤波、去噪等预处理,及综合优化的云检测处理;在此基础上,利用改进的正交匹配小波算法和优化的遗传—模糊聚类算法,实现了西北太平洋地区台风云系特征的提取。这两种算法提取出的台风云系特征参数简单,用于热带气旋(Tropical Cyclone,TC)的监测预警中可提高效率,适用于台风预报预警监测的业务化应用。

The North-West Pacific Ocean is one of the places that have the maximal frequency of the Tropical Cyclones（TC） and the highest intensity of TCs in the world,especially to which intensity level up to typhoon.Typhoon is one of the disaster weather systems of serious influence the mankind produce living,while our country is one of the stricken serious country that is subjected to it.Typhoon brings the very great threaten and disasters to industry and agriculture production,traffic and transportation,the safe of life and property of people in cities of near southeast of sea in our country.It is significant to investigate the characteristics of TCs in the North-West Pacific Ocean.With the development of detecting technique,nephogram became an important tool for monitoring tropical cyclone and typhoon in the tropical ocean,effective classification of image data and the segmentation of TC cloud are key steps for subsequent image feature extraction.Numerous scientific research had been launching into the area,compositive efficient algorithms were still in dire need.In the paper,the compositive orthogonal—matching pursuit algorithm,which was proved the efficiency in practice,and the compositive GA-FCM（Genetic Algorithms and Fuzzy C-Means algorithm） were introduced to segment the typhoon cloud systems.Although the veracity is needed to be improved,the algorithm could efficiency segment the main typhoon cloud systems,and can satisfy the forecast and monitoring of TCs well in the meteorological applications.