基于面向对象方法的白天雾检测

通过遥感技术进行雾检测的难点在于云和雾的分离,传统的基于像素的雾检测方法只考虑到云和雾的光谱信息,但是云和雾的光谱信息极为相似,不易区分。本文提出一种基于面向对象的白天雾检测方法,首先从遥感影像中提取云雾目标,然后对云雾目标进行分割,最后选取云雾分离的特征参数,完成对雾区的检测。试验结果显示,相对于传统的基于像素光谱信息的多层次阈值法,本文提出的基于对象的雾检测方法相对于传统方法在检测效果上有较大的提高。

Daytime Radiation Fog Detection Based on Object-oriented Method

The difficulty in daytime fog detection with remote sensing technology lies in distinguishing fog and cloud.Most of the traditional fog detection methods based on pixel-level are only using the spectral information of remote sensing data and due to similar spectral characteristics of fog and cloud,these methods have low precision in fog detection.In this paper a new method based on object-oriented technique is presented.Firstly the non-fog and non-cloud parts are removed from remote sensing image,and then the remainder parts are segmented into objects,finally several feature parameters are selected to determine whether each object is fog or cloud.Experiments indicate that this method improves the precision of day fog detection compared to the traditional methods.