摘要
高分遥感影像中云阴影的存在严重影响了遥感影像的质量,为了减小这一影响,提出了利用云定位的云阴影自动检测方法。首先利用云的高亮度特征,使用光谱特征检测法进行云的自动检测;然后利用阴影亮度低、归一化蓝色分量高的特性,构建云阴影的自动初检测;再根据光源、云及其阴影之间的几何光照模型,沿着光照方向,利用云的检测结果对云阴影进一步定位搜寻,利用形状和大小快速寻找到对应的云阴影区域,并根据云检测结果对云阴影的大小和形状进行优化修正,得到云阴影的最终检测结果。实验表明此方法能有效地提高云阴影检测的准确率,可有效排除水体、山体等低亮度非阴影区域的干扰。
The existence of cloud shadow in high-resolution remote sensing image seriously affects the impact of image quality.To reduce the impact,we proposed a cloud shadow automatic detection method using cloud localization.Firstly,we used the spectral feature detection method to detect the cloud automatically.Then,we used the high blue component and the low brightness feature of shadows to construct the automatic initial detection of the cloud shadow.According to the geometric illumination model between the light source,the cloud and its shadow,we used the cloud shadow results to search the cloud shadows in the illumination direction.Finally,based on the cloud shape and size,it would be easily and quickly to find the corresponding cloud shadow area,and to optimize the cloud shadow area.In addition,using the morphological processing method to fill small holes in cloud shadow area could improve the completeness.The experimental results indicate that this method can effectively improve the correctness of cloud shadow detection result.Compared with other method using shadow features directly,the proposed method can efficiently exclude other non-cloud-shadow such as water and hill with low illumination.
出处
《地理空间信息》
2020年第4期70-73,I0007,共5页
Geospatial Information
基金
测绘遥感信息工程国家重点实验室开放基金资助项目(18R04)
地理国情监测国家测绘地理信息局重点实验室开放基金资助项目(2017NGCM07)
湖北省教育厅科学研究计划资助项目(Q20181317)。
关键词
高分遥感影像
云检测
阴影检测
云阴影自动检测
云定位
high-resolution remote sensing image
cloud detection
shadow detection
cloud shadow automatic detection
cloud location
