海岸水边线提取方法在GF-2卫星影像中的适应性研究

准确提取海岸水边线是海岸线变迁研究中必不可少的基本工作。本文以海南省陵水黎族自治县2016年的GF-2影像为数据源，在大气校正和正射校正等预处理的基础上，结合数学形态学处理等方法，采用NDWI (Normalized Difference Water Index)阈值法、ISOADTA (Iterative Self-Organizing Data Analysis Technique Algorithm)、SVM (Support Vector Machine)和面向对象法（Object-Oriented Classification,OOC）等4种不同方法来对海岸水边线进行提取，并对其提取结果进行定性和定量评估，进而分析比较了这4种方法在提取人工海岸、砂质海岸和基岩海岸等3种不同海岸类型的水边线时的提取效果和提取精度。结果表明，SVM方法在3种海岸水边线上的总体提取效果最好，总体提取精度最高。

Adaptability of coastal waterline extraction methods in GF-2 satellite imagery

It is an essential work to extract the coastal waterline accurately in the study of coastline change. In this paper, one GF-2 image of Lingshui Autonomous County in Hainan Province in 2016 was taken as the data source. On the basis of atmospheric correction and orthorectification, combined with mathematical morphology processing and other methods, NDWI (Normalized Difference Water Index) Threshold Segmentation, ISODATA (Iterative Self-Organizing Data Analysis Technique Algorithm), SVM (Support Vector Machine) and OOC (Object-Oriented Classification), were used to extract the coastal waterlines, and the extraction results were qualitatively and quantitatively evaluated. Then, the extraction effects and extraction accuracy of these four methods in extracting the coastal waterlines of three different coastal types, namely artificial coast, bedrock coast and sandy coast, are analyzed. The results show that the SVM method has the best overall extraction effect and the highest overall extraction accuracy on the three types of coastal waterlines.