摘要
在高分辨的遥感卫星图像中常常会存在云层遮挡的情况,这对进行一些图像分类或者目标识别任务造成了很大的干扰,因此如何快速有效地判断遥感图像中是否存在云遮挡是一个关键性的技术。大部分的检测算法都是基于阈值分割来进行判断的,这些算法常常面临着适应性不强和检测速度慢的问题。论文提出了一种基于改进的最大响应滤波器(Maximum Response Filter,MR)的快速云检测算法,通过提取的特征和事先训练好的分类器,能快速而有效地进行云区检测,并在我国高分1号卫星图像上进行实验,得到了令人满意的结果,证明了该方法的有效性。
The cloud usually exists in high resolution remote sensing satellite images,which makes it hard to implement the tasks of image classification and object detection.Therefore,fast and precise cloud detection has become the key point that we need to solve.However,most methods in cloud detection are based on threshold segmentation that usually are inefficient and high time-consuming.In this paper,a fast cloud detection based on improved Maximum Response Filter(MR)is proposed.The experiments on GF-1 satellite remote sensing image show that a better result with low time-consuming property is obtained.
作者
关峥嵘
谭毅华
田金文
GUAN Zhengrong;TAN Yihua;TIAN Jinwen(School of Automation,Huazhong University of Science and Technology,Wuhan 430074)
出处
《计算机与数字工程》
2019年第11期2808-2811,共4页
Computer & Digital Engineering
关键词
遥感图像
最大响应滤波器
云检测
特征提取
remote sensing
maximum response filter
cloud detection
feature extraction
