基于ZY-3遥感影像的不同地貌水边线提取方法

Study on the Extraction of Waterline with Different Landforms Based on ZY-3 Remote Sensing Images

水边线的精确提取对于沿海地区的经济开发和海域的使用管理具有重要意义。以雷州半岛东北部为研究区域,利用2017年资源三号(ZY-3)卫星数据为数据源,基于不同海岸地貌特征为划分依据,运用阈值分割法、神经元网络分类法和面向对象法对多光谱数据的人工海岸、砂质海岸、淤泥质海岸和红树林海岸进行水边线提取。通过目视解译提取融合图像的海岸线为基线,将提取的水边线与基线进行定性、定量分析。研究结果表明,对于人工岸线,神经元网络分类法最优,均方根误差为6.4m;对于砂质岸线,阈值分割法最优,均方根误差为5.4m;对于淤泥质及红树林岸线,面向对象法最优,均方根误差分别为23.3m和15.2m。该研究对于不同岸线的提取具有重要的借鉴和指导意义。

Accurate extraction of waterlines is significant to the economic development of coastal areas and the management of sea areas. The northeastern of Leizhou Peninsula is taken as the research area, and the satellite data of 2017 (ZY-3) is used as the data source.Based on the geomorphological features of different coasts,the threshold segmentation method, neural network classification method, and object-oriented classification method are used to extract the artificial coastlines, sandy coastlines, silty coastlines, and mangrove coastlines with multispectral data.The coastlines extracted from fusion image by manual visual interpretation as the baseline, the extracted waterlines and baselines are qualitatively and quantitatively analyzed. The experimental results show that for the artificial shoreline, the neural network classification method is optimal with the root mean square error is 6.4m;for the sandy shoreline, the threshold segmentation method is optimal with the root mean square error is 5.4m;for muddy and mangrove shorelines,the object-oriented classification method is optimal with root mean square errors of 23.3m and 15.2m, respectively.This study has important reference and guiding significance for extraction of different shorelines.