| 摘要: |
| 海岸线是海洋的重要组成部分，对于海域管理具有重要意义。本文根据岸线位置确定原理，即将多时相瞬时水边线上边界近似作为海岸线，提出了一种基于多时相水边线的海岸线自动提取方法，该方法应用区域生长与边缘检测相结合的方法提取瞬时水边线，然后采用海岸线自动判别算法进行海岸线的提取。首先计算归一化水体指数（normalized difference water index，NDWI）图像，利用区域生长算法进行海陆分割，得到海陆分割二值图；然后利用Canny边缘检测对海陆分割结果进行边缘检测，得到瞬时水边线；最后利用海岸线自动判别算法得到多时相瞬时水边线上界作为海岸线。本文以马六甲海峡部分海岸作为研究区域，应用多期Landsat 8遥感影像进行了海岸线提取和分析。为验证海岸线提取精度，对获取时间接近高潮时的一景哨兵二号影像进行目视解译得到人工解译的海岸线，与本文提取海岸线进行对比。结果显示，本文算法得到的海岸线平均偏移量和均方根误差分别为21.01 m和21.96 m，能够满足30 m分辨率遥感图像的精度要求，相比采用单景影像提取海岸线，精度有明显提高。 |
| 关键词: 海岸线 多时相 遥感 马六甲海峡 |
| DOI：10.11759/hykx20201108004 |
| 分类号:TP79 |
| 基金项目:国家重点研发计划项目（2017YFC1405600）；国家自然科学基金项目（41776182）；山东省自然科学基金-面上项目（ZR2019MD023） |
|
| Automatic coastline extraction method based on multitemporal remote sensing data |
|
SHENG Hui, ZHANG Chi, WAN Jian-hua
|
|
College of Ocean and Space Information, China University of Petroleum (East China), Qingdao 266580, China
|
| Abstract: |
| A coastline is an essential part of the ocean and is considerably important to the management of sea areas. According to the principle of shoreline location determination, this study proposes an automatic coastline extraction method based on a multitemporal instantaneous waterline, which uses the region growth method and edge detection to extract the instantaneous waterline, and an automatic coastline discrimination algorithm to extract the coastline. First, the NDWI image is calculated, a region growing algorithm is used to segment the land and sea, and a binary image is obtained. Second, the Canny operator is used to detect the edge of the land-sea segmentation result to obtain the instantaneous waterline. Finally, the multitemporal instantaneous waterline boundary is obtained as the coastline using the automatic coastline discrimination algorithm. In this study, a section of the coast of the Malacca Strait is taken as the research area, and the coastline is extracted and analyzed using multiperiod Landsat 8 remote sensing images. To verify the accuracy of the coastline extraction, a Sentinel-2 scene image when the acquisition time was close to high tide is visually interpreted, and the artificial interpretation coastline is obtained, which is compared with the coastline extracted for this study. Results show that the average offset and root mean square error of the coastline obtained by the algorithm are 21.01 and 21.96 m, respectively, which meet the accuracy requirements of 30-m resolution remote sensing images, and the accuracy is significantly improved compared to that of single-scene image extraction. |
| Key words: coastline multitemporal remote sensing Strait of Malacca |
