基于极化散射特性的神经网络海岸带信息提取

详细信息    查看全文 | 推荐本文 |

英文篇名：The Coastal Zone Information Extraction Based on the Neural Network Method With Polarization Characteristics 作者：梁超 ; 曾韬 ; 邹亚荣 英文作者：LIANG Chao;ZENG Tao;ZOU Ya-rong;National Satellite Ocean Application Service;Key Laboratory of Space Ocean Remote Sensing and Application,SOA; 关键词：SAR ; 极化分解 ; BP人工神经网络 ; 海岸带 英文关键词：SAR;;polarization decomposition;;BP ANN;;coastal zone 中文刊名：HBHH 英文刊名：Advances in Marine Science 机构：国家卫星海洋应用中心;国家海洋局空间海洋遥感与应用研究重点实验室; 出版日期：2017-04-15 出版单位：海洋科学进展 年：2017 期：v.35 基金：国家自然科学基金项目——随机粗糙海面的溢油极化SAR特征谱构建及精细提取研究(41376183);; 南北极环境综合考察与评估专项——南极地区环境遥感考察(CHINARE2015-02-04) 语种：中文; 页：HBHH201702014 页数：6 CN：02 ISSN：37-1387/P 分类号：152-157

摘要

应用遥感手段开展海岛海岸带监测为海岸带资源开发利用提供科学的信息支撑。采用Radarsat-2数据,开展Cloude极化,获得分解参数,进行海岸带地物特性极化参数特性研究,在此基础上,以辽宁鲅鱼圈作为研究区域,运用神经网络方法开展海岸带信息分类提取研究。结果表明:极化目标分解理论对海岸带信息提取具有一定的应用潜力,采用基于H/α的分类方法能较好地区分单次散射的特征地物,但对于偶次散射和体散射的混合体,仅从极化特征参数还难以区分;综合利用极化散射特性及神经网络分类方法则可以有效进行分类,采用基于SPOT5数据的"我国近海海洋综合调查与评价专项"遥感调查成果为验证标准,精度达到88.5%。分类精度与训练样本有关,此外,海岸带区域地物分布往往较为复杂,对于复杂地物的散射机制研制,是下一步研究工作的重点。
        Application of remote sensing data in the monitoring of island and coastal zone can provide scientific information support for the development and utilization of coastal zone resources.Radarsat-2data is used to carry out the coastal information extraction with Cloude polarization decomposition theory.The results show the application potential of Cloude polarization decomposition theory in the coastal information extraction.The objects of single scattering characteristics can be easier distinguished based on H/αclassification method,but it is difficult to distinguish the targets of even scattering and volume scattering mechanism.The neural network method together with the polarization characteristics can be used for the coastal zone classification,and the accuracy is 88.5%,which is related to the training samples.The distribution of surface features in the coastal zone is complicated,which is the focus of the future research work.

引文

[1]WANG C Y.Coastal land covers classification of remote sensing images based on data minning technology[D].Qingdao:Ocean University of China,2009.王常颖.基于数据挖掘的遥感影像海岸带地物分类方法研究[D].青岛:中国海洋大学,2009.
    [2]HE H J,WANG W,LIU X G.Land type classification of coastal zone based on decision tree model[J].Geography and Geo-Information Science,2008,24(5):25-28.何厚军,王文,刘学工.基于决策树模型的海岸带分类方法研究[J].地理与地理信息科学,2008,24(5):25-28.
    [3]REN G B.The study of semi-supervised classification technology in remote sensing image processing[D].Qingdao:Ocean University of China,2010.任广波.基于半监督学习的遥感影像分类技术研究[D].青岛:中国海洋大学,2010.
    [4]FU J,GU D Q.Extracting method in tidal flat based on decision fusion of spectrum and texture[J].Marine Environmental Science,2008,27(5):405-408.付军,谷东起.基于光谱与纹理特征融合的滩涂信息提取方法[J].海洋环境科学,2008,27(5):405-408.
    [5]LIU Y F,ZHANG J,MA Y,et al.Incorporating Geoscience knowledge into land use/cover classification for remote sensing image of coastal zone[J].Advances in Marine Science,2010,28(2):193-202.刘艳芬,张杰,马毅,等.融合地学知识的海岸带遥感图像土地利用/覆盖分类研究[J].海洋科学进展,2010,28(2):193-202.
    [6]CHEN J Y,PAN D L,MAO Z H.Optimal segmentation problem of simple object in high resolution coastal zone remote sensing image[J].Science in China:Series D Earth Sciences,2006,36(11):1044-1051.陈建裕,潘德炉,毛志华.高分辨率海岸带遥感影像中简单地物的最优分割问题[J].中国科学:D辑地球科学,2006,36(11):1044-1051.
    [7]ZHANG X,DENG K Z,FAN H D,et al.PolSAR SVM supervised classification method combining with polarimetric target decomposition[J].Application Research of Computers,2013,30(1):295-298.张祥,邓喀中,范洪冬,等.基于目标分解的极化SAR图像SVM监督分类[J].计算机应用,2013,30(1):295-298.
    [8]WANG C,ZHANG H,CHEN X.Full polarization synthetic aperture radar image processing[M].Beijing:Science Press,2008.王超,张红,陈曦.全极化合成孔径雷达图像处理[M].北京:科学出版社,2008.
    [9]ZOU Y R,LIANG C,ZENG T.Oil spill identification using SVM based on polarization parameters[J].Marine Study,2013,31(3):71-76.邹亚荣,梁超,曾韬.基于极化参数的SVM海上溢油识别[J].海洋学研究,2013,31(3):71-76.
    [10]LEE J S,POTTIER E.Polarimetric radar imaging:from basics to applications[M].Boca Raton:CRC Press,2009:267-268.
    [11]LI H D,GUAN D X,YUAN F H,et al.Modeling canopy transpiration of young poplar trees(Populus×euramericana cv.N3016)based on Back Propagation Artificial Neural Network[J].2015,35(12):1-13.李辉东,关德新,袁凤辉,等.BP人工神经网络的杨树林冠蒸腾[J].生态学报,2015,35(12):1-13.