| 摘要: |
| 潟湖是水陆交接的地貌体，结合遥感手段实现水陆的准确、快速分割对于潟湖岸线长度和面积信息提取具有重要意义.以山东半岛的海岸潟湖为研究对象，利用7景2010年左右覆盖研究区域的Landsat TM图像，在图像几何校正和信息增强的基础上，采用敏感波段组合的阈值分割方法，通过对比归一化植被指数（NDVI）、归一化水体指数（NDWI）和改进的归一化差异水体指数（MNDWI）方法处理潟湖边界信息的敏感性，确定了基于NDWI的潟湖水陆分割方法，并通过栅矢转换完成了潟湖边界范围的自动提取，为了避免由于栅格精度限制而导致的边界信息提取结果不圆滑，采用样条函数法的插补算法进行平滑处理.同时，为了验证潟湖信息自动提取的精度，提出了一种基于相同地理位置的斑块精度验证方法，利用人工目视解译的结果，从潟湖的形状、岸线长度和面积3个角度评价自动提取的结果，该方法可验证潟湖边界自动提取结果与真实潟湖边界的形状相似度，并对相似度较好的潟湖信息提取结果进行精度分析. |
| 关键词: 海洋遥感学 潟湖 信息提取 精度验证 NDWI Landsat TM图像 |
| DOI：10.3969/J.ISSN.2095-4972.2016.03.015 |
|
| 基金项目:中央级公益性科研院所基本科研业务费专项资金资助项目（2015T03, 2014G16）;中欧国际合作龙计划三期资助项目（ID10470） |
|
| Shoreline remote sensing extraction and accuracy verification method of coastal lagoon |
| SUN Weifu,ZHANG Jie,MA Yi,XIA Dongxing |
| (First Institute of Oceanography， SOA， Qingdao 266061，China) |
| Abstract: |
| Lagoon is special landform between land and water. It is significant to segment land and water accuracy and fast， extracting lagoon shoreline length and area information by means of remote sensing. In this paper， Shandong Peninsula lagoons are taken as the research objects， based on the 7 senses of Landsat TM images of 2010， we contrast with the normalized difference vegetation index （NDVI）， normalized difference water index （NDWI） and modified normalized difference water index （MNDWI） method that sensitive to the lagoon boundary information. NDWI method was used to segment water and land in the lagoon. We convert the raster to vector to complete the shoreline boundary automatic extraction of the lagoon. In order to avoid the rough results of boundary information extraction due to the limitation of grids accuracy， the method of spline function interpolation is used to smooth the edge information. While， using the results of artificial interpretation， this paper presents a feature accuracy verification method based on the same location to verify the accuracy of information automatic extraction of the lagoons according to the shoreline length， lagoon shape and area. The method can verify the shape similarity between automatically extracted and real lagoon boundary， and then verify the accuracy of the result according to the better similarity feature． |
| Key words: marine remote lagoon information extraction accuracy verification NDWI landsat TM image |
