联合光谱和纹理特征的支持向量机滩涂分类方法研究  

Research on joint spectral and texture features SVM tidal classification and extraction algorithm

作　　者：赵亮 孟立轲[2] 张艳[3] 冯伍法 张继领 

机构地区：[1]61175部队 [2]96656部队 [3]信息工程大学地理空间信息学院 [4]68310部队

出　　处：《测绘工程》2016年第1期43-46,共4页Engineering of Surveying and Mapping

摘　　要：滩涂作为海岸带的重要组成部分,是重要的土地资源。针对遥感影像滩涂分类的提取,文中提出一种联合光谱和纹理特征支持向量机(SVM)滩涂分类的方法。首先介绍纹理特征影像获取方法,通过灰度共生矩阵分析得到滩涂纹理特征影像;然后将光谱影像与纹理影像叠加形成一幅多维特征影像,用SVM分类算法中的OAR分类器进行分类实验,对分类结果进行实验分析。实验结果表明,该算法对提高海岸带地理信息获取能力,提升海洋遥感测绘信息化保障水平有积极意义。As a main part of the beach coastal zone,tidal land is an important land resource.In this paper,study is made on remote sensing image tidal land classification and extraction technology,and a tidal land SVM classification method combining the spectral and texture features is proposed.Firstly,the texture image acquisition method is introduced,then tidal land texture image is obtained by GLCM analysis,subsequently a multi-dimensional feature image is generated by superimposing the spectral image and the texture image together,finally the SVM classification algorithm with OAR classifier is chosen for classification.Experimental results demonstrate the algorithm is significant for enhancing coastal geographic information access capabilities and lifting the marine remote sensing surveying and mapping level.

关 键 词：光谱特征 纹理特征 支持向量机 滩涂分类 

分 类 号：P237[天文地球—摄影测量与遥感] P931.1[天文地球—测绘科学与技术]