【摘要】 针对气象学科基础研究中的雪花形状分类问题,文章提出一种基于BP神经网络的雪花形状分类方法。首先对雪花图像进行图像预处理并提取雪花的轮廓特性;在此基础上利用雪花轮廓得到雪花的纵横轴比、矩形度、周长凹凸比、面积凹凸比、形状参数以及致密度6种形状特征参数,采取BP神经网络设计分类器。实验表明,该分类器识别率可以达到91.67%,能够为后续研究雪花物理结构与人工干预降雪之间的关系提供可靠数据支撑。更多还原

【Abstract】 According to the snowflake shape classification in the basic research of meteorological disciplines, this paper proposes a snowflake shape classification method based on BP neural network. First, the image of the snowflake image is preprocessed and the outline characteristics of the snowflake are extracted. On the basis of the snowflake profile, the aspect ratio, the squareness, the perimeter of the snowflake, the area roughness ratio, the shape parameters and the density of the snowflake are obtained. Parameters, adopt BP neural network design classifier. Experiments show that the recognition rate of this classifier can reach 91.67%, which can provide reliable data support for the subsequent study on the relationship between snowflake physical structure and artificial intervention snowfall.更多还原