基于Photoshop的Landsat-8影像云及云影提取方法

Extraction Method of Cloud and Cloud Shadow Decision Based on Photoshop

云及云影不可避免地作为无信息或弱信息区存在于遥感影像中,成为制约遥感影像应用深度和广度的一大问题。云及云影区的精确定位是多源、多时相遥感影像云及云影区信息重建的基础。薄、厚云光谱与空间形态特征各异,难以仅通过简单的特征实现云及云影的自动化、高精度提取。Phtoshop(PS)作为功能强大且最为常用的图像处理软件,其强大的色彩分析与选区功能使得图像像素信息分析较为便捷。在对云、冰雪及其他地类光谱特性分析的基础上,首先选择能够有效区分云、冰雪及其他地类的红、短波红外及热红外波段合成彩色图像,借助于色彩相似性容差控制下的PS像素选区选取与"种子点"空间增长功能,实现云信息的有效提取;其次,通过云选区向阴影方向的移动对云影的初定位,结合近红外波段上灰度相似性容差控制的PS像素选区选取与"种子点"空间增长功能,实现云影信息的提取。对Landsat-8的实验结果表明,利用PS能够有效提取出云及云影信息,其中云信息具有优于96%的提取精度,云影信息提取精度超过90%。云及云影信息的PS有效提取方法为自动化提取算法研发乃至其他地类信息的提取提供了可借鉴的思路。

Clouds and its shadow is always the weak and even uninformative area inevitably exists in the remote sensing images,and greatly restrict the field of remote sensing applications.Precise positioning cloud and its shadow area is an important step for information reconstruction of these area by using multi-source or multi-temporal remote sensing images.Thin and thick clouds always have different spectral and spatial morphological characteristics,and it is difficult to achieve automation and high-precision extraction through some simple features.Phtoshop（PS）as a powerful and most commonly used image processing software,its powerful color analysis and selected-area operation make image pixel information analysis more convenient.Based on the analysis of the spectral characteristics for the cloud,snow,and other class,and PS-based cloud and its shadow information extraction method was proposed：firstly,red,short-wave infrared and thermal infrared band was select to synthesize color image for distinguish cloud,snow,and other class by using of the pixel selection and ＂seed point＂ expansion function with the color similarity tolerance;secondly,using the movement of the cloud selected-area to achieve the initial position,and then the near-infrared band was selected to extract cloud shadow by using of pixel selection and ＂seed point＂ expansion function with the intensity similarity tolerance.The Landsat-8experimental results show that the PS can effectively extract the cloud and its shadow information,which has a cloud information extraction accuracy better than 96%,and the accuracy of cloud shadow information extraction better than 90%.The efficient PS clouds and its information extraction method provide some useful ideas for the automated extraction algorithm development as well as other types of information extraction.