多光谱与纹理特征结合的WorldView-2影像海岸线提取方法研究——以马来西亚棉花岛为例

Research on coastline extraction method of WorldView-2 image combined with multi-spectral and texture features:Take the Cotton Island of Malaysia as an example

海岸线是海岛海岸带研究的基础要素,对于海岸带资源开发利用具有重要意义。遥感技术凭借其宏观、多光谱且周期短等特点成为快速提取海岸线的重要技术手段。本文采用高分辨率遥感数据WorldView-2对海岸线两侧地物的表观反射率进行分析,并开展特征分析,构建了海岸线提取指数。在此基础上,进一步结合纹理特征提取方法,成功提取了马来西亚棉花岛的海岸线。基于线目标匹配法对提取结果进行精度评价,其中长度误差为-1.38%,在缓冲半径为6个像素单位时,完整度为96.96%,正确度为95.37%。结果表明,用该方法提取的海岸线精度较高,可满足遥感调查的需要。

Coastline is a fundamental element of the island and coastal zone research,which is of great significance to the exploitation and utilization of coastal resources.Remote sensing technology has become an important technical means for quickly extracting coastlines due to its macroscopic,multi-spectral and short cycle characteristics.In this study,the apparent reflectance of ground objects on both sides of the coastline was analyzed by using high-resolution remote sensing data WorldView-2,and the feature analysis was carried out to construct the coastline extraction index.On this basis,the shoreline of the Cotton Island,Malaysia was successfully extracted by further combining the texture feature extraction method.Then the accuracy of the extraction results was evaluated based on the lines matching method.The length error is-1.38%.When the buffer radius is 6 pixels,the integrity is 96.96% and the accuracy is 95.37%.The results show that this method could effectively extract the coastline and meet the demand for remote sensing survey.