基于OTSU在MODIS数据中的云检测应用

Application of Cloud Detection in MODIS Data Based on OTSU

MODIS数据信息量大,具有丰富的研究价值,为了更好地对MODIS数据进行利用,就需要进行云检测的研究。论文针对MODIS数据的特点,通过平均方差替代均值的计算方式,改良传统的OTSU算法,从而得出对应阈值,融入到多波段阈值法中,将遥感影像分割为云和非云两部分,使得边缘地区的薄云识别的效果更加明显,多学科之间相结合实现了云检测的可视化和自动化,提高了云检测整体精度。实验表明,在黄渤海地区该方法较固定阈值法效果提升明显,更能够适应不同时相的云,并且总体精度达到94%以上。

MODIS data has a large amount of information and rich research value.In order to make better use of MODIS data,it is necessary to conduct cloud detection research.In view of the characteristics of MODIS data,this paper improves the traditional OTSU algorithm by replacing the mean value with the average variance to obtain the corresponding threshold,which is integrated in-to the multi-band threshold method,and the remote sensing image is divided into two parts of cloud and non-cloud.The effect of thin cloud recognition in edge areas is more obvious.The combination of multiple disciplines realizes the visualization and automa-tion of cloud detection,and improves the overall accuracy of cloud detection.Experiments show that in the Yellow Sea and Bohai Sea,this method has a significant improvement over the fixed threshold method,and is more adaptable to clouds of different time phases,and the overall accuracy is more than 94%.