基于Q-OTSU算法高分三号水体快速提取方法研究——以河南鹤壁洪涝灾害监测为例

Research on rapid extraction method of water based on Q-OTSU algorithm using domestic GF-3 image: A case of flood disaster monitoring in Hebi,Henan Province

最大类间方差法(OTSU)算法是一种确定图像二值化分割阈值算法,其用于合成孔径雷达(SAR)影像中进行地物目标快速提取有突出的优势。本文提出一种优化的OTSU算法(Quick-OTSU,Q-OTSU),可以快速提取SAR影像中的水体信息,满足洪涝灾害快速应急需求。2021年7月,河南省遭受持续性强降雨,引发特大洪涝灾害,本文采用多景国产高分三号卫星(GF-3)不同时相SAR数据,采用Q-OTSU阈值分割法快速提取河南省鹤壁市受灾区域水体信息,同时利用同期哨兵二号(Sentinel-2B)影像进行精度验证,分析受灾区域灾害情况。在获取影像24 h内就完成了水体信息提取,精度达到96.7%。研究结果表明:本文提出的Q-OTSU算法应用于国产高分三号数据水体提取中,精度高,可高效快速进行洪水灾情评估,能够为政府和相关部门及时、精确地掌握洪灾信息,科学防灾救灾提供了决策依据。

OTSU(maximum between-cluster variance)is an algorithm to determine the threshold of image binary segmentation. It has outstanding advantages in ground object fast extraction in synthetic aperture radar(SAR)images. This paper presents an optimized OTSU algorithm(Quick-OTSU,Q-OTSU),which can quickly extract the water information in SAR images,so as to meet the rapid emergency needs of flood disasters. In July 2021,Henan Province suffered from continuous heavy rainfall,causing a catastrophic flood disaster. In this paper,the multi-scene domestic GF-3 satellite(GF-3)SAR data in different time phase are used to quickly extract the water information of the affected area in Hebi City,Henan Province by using the Q-OTSU threshold segmentation method,and the sentinel-2B image in the same period is used to verify the accuracy and analyze the disaster situation in the affected area. The water information extraction was completed within 24 hours after obtaining the image,and the accuracy reaches 96.7%. The results show that the Q-OTSU algorithm proposed in this paper is applied to the water information extraction of domestic GF-3 data,has high accuracy,which can carry out flood disaster assessment effectively,and provide decision-making basis for the government and relevant departments to grasp flood information in time and accurately and scientific disaster prevention and relief.