论文部分内容阅读
Landsat 8 OLI影像已成为重要的数据源,但受云及云影的影响较大,降低了数据的可用性,因此,快速识别云及云影,为后续的数据恢复有着积极的应用价值。通过构建云指数(CI)、归一化暗像元指数(NDPI)和比值阴影指数(RSI),采用阈值法和方位角搜索法,以两景Landsat 8 OLI影像(一景试验影像,另一景验证影像)为例进行云及云影检测。每类随机选取200个样点进行精度分析,结果表明:CI可快速区分OLI影像中的云区与非云区,厚云样本点正确识别率达到99%;NDPI与归一化植被指数(NDVI)构建的比值阴影指数RSI放大了水体、云影与其他阴影间的差异,更便于区分;方位搜索合理设置搜索方位角和搜索距离,简化了云影与云的相对关系模型,可准确区分水体与云影,两者的正确识别率都超过93%,弥补了阈值法的局限性。本方法可行快捷,为OLI影像的后续应用提供了基础,可有效提高其利用精度。
Landsat 8 OLI image has become an important data source, but it is greatly affected by clouds and cloud images, reducing the availability of data. Therefore, quickly identifying cloud and cloud images has positive application value for subsequent data recovery. By constructing cloud index (CI), normalizing dark pixel index (NDPI) and ratio shadow index (RSI), using threshold method and azimuth search method, using Landsat 8 OLI images King verification video) as an example for cloud and cloud shadow detection. The results show that: CI can quickly distinguish between cloud and non-cloud regions in OLI images, and the correct recognition rate of thick cloud samples reaches 99%; NDPI and normalized vegetation index (NDVI ) To construct the ratio shadow index RSI to amplify the difference between the water body, the cloud shadow and other shades and make it easier to distinguish; the azimuth search sets the search azimuth and the search distance reasonably, simplifies the relative relation model between the cloud shadow and the cloud, And the shadow of the cloud, both the correct recognition rate of more than 93%, to make up for the limitations of the threshold method. The method is feasible and fast, which provides the foundation for the subsequent application of OLI images, which can effectively improve the utilization accuracy.