基于Landsat8数据的近海养殖区自动提取方法研究  

Research on automatic extraction method for coastal aquaculture area using Landsat8 data

作　　者：武易天 陈甫 马勇 刘建波 李信鹏 WU Yitian;CHEN Fu;MA Yong;LIU Jianbo;LI Xinpeng(Institute of Remote Sensing and Digital Earth,CAS,Beijing 100101,China;University of Chinese Academy of Science,Beijing 100049,China)

机构地区：[1]中国科学院遥感与数字地球研究所,北京100101 [2]中国科学院大学,北京100049

出　　处：《国土资源遥感》2018年第3期96-105,共10页Remote Sensing for Land & Resources

基　　金：中国科学院135突破项目资助

摘　　要：利用遥感影像进行近海养殖区提取,是近海渔业资源监管的一种有效手段。然而由于近海养殖区受海水叶绿素和悬浮泥沙浓度差异分布的影响,海域水色背景往往较为复杂,使得传统近海养殖区提取方法很难取得理想效果。为了准确提取复杂水色背景下的近海养殖区,提出一种结合光谱和纹理信息的养殖区自动提取方法。首先,利用基于正交子空间投影的约束能量最小化(orthogonal subspace projection-weighted constrained energy minimization,OWCEM)方法增强近海养殖区信息;然后,利用近海养殖区的空间纹理信息,使用局部标准差自适应分割法(standard deviation adaptive segmentation,SDAS)对影像进行划分;最后,基于分割图像对近海养殖区进行自动提取。选取山东桑沟湾养殖海域和福建三都澳养殖海域为实验区,采用Landsat8数据对所提方法进行近海养殖区提取实验,结果表明:在复杂水色背景下,该方法能够准确快速地识别出近海养殖区的分布情况,在2个实验区都能达到93%以上的准确率,为近海养殖区自动提取提供了一种新的有效方法。During coastal resource monitoring,it is an effective way to extract aquaculture region using remote sensing data,whereas the water color in coastal region is complexly influenced by the distribution difference of chlorophyll-a and total suspended sediment concentration.And it would be difficult to accurately extract the aquaculture region with complex background using traditional methods.In view of the above problem,the authors proposed an algorithm for automatic coastal aquaculture area extraction combined with spectral and spatial information of aquaculture.Firstly,orthogonal subspace projection-weighted constrained energy minization method(OWCEM)was used to enhance the information of coastal aquaculture area.Secondly,by using the spatial texture information of the coastal aquaculture area,standard deviation adaptive segmentation(SDAS)method was used to automatically extract the cultivation area.In order to verify the accuracy of the proposed algorithm,the authors selected Sanggou Bay in Shandong and Sanduao Bay in Fujian as test regions and conducted the area extraction using Landsat8 data.The experimental results show that the proposed method can rapidly and accurately identify the distribution of coastal aquaculture area in complex background color and can reach about 93%accuracy rate with a low missing rate.The method could provide a new and effective means for automatic extraction of offshore aquaculture area.

关 键 词：近海养殖区提取 标准差自适应分割(SDAS) 复杂背景 基于正交子空间投影的约束能量最小化(OWCEM) 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]