论文部分内容阅读
为了实现冰雹暴雨天气的识别与分类,提出了一种基于雷达反射率图像特征的自动识别方法.对雷达回波反射率图像中冰雹回波区域和暴雨回波区域的图像特征进行提取,通过分析冰雹暴雨间单一特征的差异性和不同特征之间的分类互补性,确定了识别冰雹暴雨的有效图像特征(包括强度特征和纹理特征).将提取出的样本有效特征与探空数据(0℃和~20℃温度层高度)结合,利用粗糙集理论进行数据挖掘,进而建立了冰雹暴雨天气的客观识别模型.通过对362个测试样本的测试与统计,冰雹击中率达到93.29%,暴雨的击中率达到89.27%,并且两者均具有较低的误警率.实验结果与传统PUP系统比较,表明利用雷达反射率图像特征实现对冰雹暴雨天气的识别与分类具有较好的效果.
In order to identify and classify hail rainstorm weather, an automatic recognition method based on radar reflectance image features is proposed.It extracts the image features of hail echo area and rainstorm echo area in radar echo reflectivity image, Hail storm and rainstorm, and the classification complementarity among different features, the effective image features (including intensity and texture features) to identify the hail rainstorm are identified.The effective features of the extracted samples and the sounding data (0 ℃ And ~ 20 ℃ temperature layer height), the rough set theory was used to do data mining, and then an objective identification model of hail storm weather was established.Through the test and statistics of 362 test samples, the hail hit rate reached 93.29% The hit rate was 89.27%, and both had lower false alarm rate.The experimental results compared with the traditional PUP system show that using the radar reflectivity image features to achieve the hail storm weather identification and classification has a good effect.