基于ONDPI的海岸养殖池塘遥感影像提取研究  

Research on Remote Sensing Image Extraction of Coastal Aquaculture Pond Based on ONDPI

作　　者：裴亮[1] 王金鑫 屈慧慧 桑学锋[2] PEI Liang;WANG Jinxin;QU Huihui;SANG Xuefeng(School of Geomatics,Liaoning Technology University,Fuxin 123000,China;Department of Water Resources,China Institute of Water Resources and Hydropower Research,Beijing 100038,China)

机构地区：[1]辽宁工程技术大学测绘与地理科学学院,辽宁阜新123000 [2]中国水利水电科学研究院水资源研究所,北京100038

出　　处：《海洋测绘》2020年第5期40-44,共5页Hydrographic Surveying and Charting

基　　金：国家自然科学基金(51679253);国家杰出青年科学基金(51625904)。

摘　　要：以Sentinel-2、Landsat-8卫星影像为数据源,以海岸养殖池塘为研究对象,针对养殖池塘的特点,提出了基于归一化差异池塘指数(NDPI)的面向对象结合法(ONDPI)。并以天津市某一海岸养殖区为例,与传统的水体指数结合直方图阈值分割法进行对比实验。实验表明ONDPI能够对遥感影像中的海岸养殖池塘进行有效提取,提取精度达到95%以上,优于其他方法,能够为海岸池塘养殖的规划管理提供参考数据。In this paper,Sentinel-2 and Landsat-8 satellite images are used as data sources,and coastal aquaculture ponds are used as research objects.According to the characteristics of aquaculture ponds,an object-oriented combination method(ONDPI)based on Normalized Difference Pond Index(NDPI)is proposed.Taking a certain area of Tianjin as an example,the experiment is compared with the traditional water body index combined with the histogram threshold segmentation method.Experiments show that ONDPI can effectively extract coastal aquaculture ponds from remote sensing images;the extraction accuracy is above 95%,which is better than other methods,it can provide reference data for planning and management of coastal pond aquaculture.

关 键 词：遥感影像 海岸养殖池塘提取 水体指数法 直方图阈值分割 归一化差异池塘指数 面向对象结合法 

分 类 号：P237[天文地球—摄影测量与遥感]