卫星云图修复方法研究与实现  

Study on realization of satellite cloud image inpainting

作　　者：蔡菲菲[1] 闫振宇[1] 郎宏山[1] Cai Feifei Yan Zhenyu Lang Hongshan(Beijing Meteorological Satellite Ground Station, National Satellite Meteorological Center, Beijing 100094, China)

机构地区：[1]国家卫星气象中心北京气象卫星地面站,北京100094

出　　处：《电子测量技术》2017年第6期109-112,共4页Electronic Measurement Technology

摘　　要：针对卫星云图中不可避免的丢线现象,基于中值滤波的基本思想,提出一种结合霍夫变换的图像修复方法。该方法在定位丢线位置的基础上,仅采集垂直方向邻域像素点,并根据窗内异常点个数自适应调整窗口尺寸进行修复。从卫星云图处理实时性要求高、主观视觉效果和客观评价标准3个方面,对传统整体变分、自适应滑动平均、结合霍夫变换的自适应中值滤波3种修复方法进行比较。通过仿真结果表明,这种根据丢线分布特征改进后的中值滤波方法更适合修复卫星云图中的丢线。An image inpainting metod combining with Hough transform was proposed to inpainting the inevitable scanline dropping in satellite cloud image,which was based on median filtering.This method on the basis of scanline dropping positioning,was only collected neighboring pixels in vertical direction,and adjusted the window size according to the number of abnormal points.The total variation,the adaptive sliding average,and the adaptive median filtering which combined with Hough transform,have been compared from the high realtime requirement,subjective vision and evaluation parameters aspects.The simulation results show that this improved median filter based on scanline dropping distribution characteristics is more suitable as a satellite cloud image inpainting method.

关 键 词：卫星云图 图像修复 中值滤波 霍夫变换 

分 类 号：TN919.8[电子电信—通信与信息系统]