文献类型：期刊文献

中文题名：气象预报中阴雨雾图像优化识别仿真

英文题名：Simulation and Optimization of Rain Fog Image in Weather Forecast

作者：王琦[1]

第一作者：王琦

机构：[1]北京联合大学师范学院

第一机构：北京联合大学师范学院

年份：2017

卷号：34

期号：7

起止页码：422-425

中文期刊名：计算机仿真

外文期刊名：Computer Simulation

收录：北大核心:【北大核心2014】；

基金：基金项目:北京地区地磁场三维可视化研究与实现(KM201311417005)

语种：中文

中文关键词：气象预报;低分辨率图像;识别

外文关键词：Weather forecast ; Image with low resolution ; Recognition

摘要：对气象预报中的阴雨雾图像的识别,能够更好的识别低分辨率图像中有用的但是没有表现出来的信息。对低分辨率图像的优化识别,需要获取图像整体色度差均值,将图像目标区域的轮廓变得光滑,完成对阴雨雾图像的识别。传统方法先提取图像的相位和幅值特征,将直方图定义为图像的特征向量,但忽略了获取图像整体色度差均值,导致图像识别精度较低。提出基于二次最大熵的模糊聚类的阴雨雾图像识别方法。利用多尺度Retinex变换理论对图像的亮度进行增强,将低分辨率图像分割为目标区域和非目标区域,获取图像整体色度差均值,给出低分辨率图像的最大判断准则,利用形态学滤波将图像目标区域的轮廓变得光滑,完成对阴雨雾低分辨率图像优化识别。实验结果表明,所提方法识别精度高,可以有效地识别出阴雨雾低分辨率图像中的重要信息。
In this paper, we propose a recognition method for cloudy, rainy, and foggy image based on fuzzy clus- tering of secondary maximum entropy. Firstly, multi-scale Retinex conversion theory was used to enhance brightness of image, and image with low resolution was divided into object region and non-object area. Then, mean value of o- verall color difference of image was acquired, and maximum judgment criteria of image with low resolution were pro- vided. Moreover, morphological filter was used to turn outline of the object area into smooth. Finally, the optimiza- tion recognition was completed. The experimental resuhs show that the method has high recognition precision and can recognize important information in cloudy, rainy, and foggy image with low resolution effectively.