摘要
高分辨率遥感影像中存在大量的云噪声,这些云层对遥感数据的后续处理和传输产生不利影响。本文针对遥感图像中云层污染问题,对多光谱影像中云层亮度、纹理、频率等特征展开分析,通过样本统计了云层与下垫面在不同方面的特征差异。基于分析的结果,利用灰度共生矩阵和Gabor滤波器的各个特征值,依据特征的可分离度选择有效的特征,并对上述特征进行云检测。对检测为云的样本,近一步采用条件边缘膨胀优化云层边缘的检测精度。实验结果证明本算法对高分辨遥感影像的云层具有较好的检测效果。
A large quantity of cloud which presents in high resolution remote sensing imagery bring disadvantages to later remote sensing imagery process and data transmission. In this paper, aim to solve the cloud-contamination problem, we analyze different features of clouds and underlying surface concerning intensity, textures and frequency via sample statistics. Based on the analysis, we use gray levd go-occurrence matrix and Gabor filter to extract the features and select better features through the separability value. We then achieve the cloud detection based on above extracted features and further refine the cloud boundary detection via conditional boundary dilation. Experiments demonstrate the effectiveness of our method in remote sensing imagery.
出处
《数字技术与应用》
2017年第9期124-127,共4页
Digital Technology & Application
关键词
遥感影像
高分辨率
云层特征提取
云检测
Remote sensing imagery
High resolution
Cloud features extraction
Cloud detection
