一种NDWI自约束遥感影像水体信息的快速检测方法

Fast water detection in remote sensing images by NDWI self-constrained method

针对传统遥感影像水体信息检测环节多、过程复杂的问题,本文提出一种基于归一化水体指数(NDWI)自约束的水体判别方法,该方法直接从影像中检测水体,能减少操作环节、提高检测速度,为水体应急监测提供快速响应的方法。通过对本方法与阈值法、分类法在不同数据中进行测试,结果显示:本方法对WorldView-2、Landsat8和Sentinel-2A影像水体检测的总体精度介于83.87%~91.53%,阈值法检测的总体精度介于85.07%~94.07%,支持向量机分类法检测的总体精度介于91.91%~97.70%;该方法对水体的检测精度不及阈值法和分类法高,但在具有较高检测精度的同时能明显简化操作,在水体监测应急响应中具有实际应用价值。

Aiming at the problems of multi-link and complex process of water detection in traditional remote sensing images,this paper proposed a water body identification method based on NDWI self-restraint,which detected water body directly from images,reduces operation links and improves detection speed.It can provide rapid response for emergency monitoring of water body.NDWI self-restraint method,threshold method and classification method were used to test in different data.The results showed that the overall accuracy of NDWI self-restraint method was between 83.87%and 91.53%for water detection of WorldView-2,Landsat8 and Sentinel-2A images,the overall accuracy of threshold method was between 85.07%and 94.07%,and the overall accuracy of support vector machine classification was between 91.91%and 97.70%.The overall accuracy of NDWI self-restraint method for water body detection was not as high as threshold method and classification method,but it can greatly simplify the operation while having a high detection accuracy,and has application value in emergency response of water detection.