基于SVM的多源信息复合的高空间分辨率遥感数据分类研究  

The High Spatial Resolution RS Image Classification Based on SVM Method with the Multi-Source Data

作　　者：张锦水[1] 何春阳[1] 潘耀忠[1] 李京[1] 

机构地区：[1]北京师范大学环境演变与自然灾害教育部重点实验室北京师范大学资源学院,北京100875

出　　处：《遥感学报》2006年第1期49-57,共9页NATIONAL REMOTE SENSING BULLETIN

基　　金：国家高技术研究发展计划863计划(2003AA131080)

摘　　要：遥感图像尤其是高空间分辨率(1—4m)遥感图像在土地利用和土地覆盖变化方面有着广阔的应用前景,传统高空间分辨率遥感图像信息提取方法存在精度和分类效率低的不足。本文提出的基于SVM的分类方法,复合光谱、纹理和结构信息等多源数据信息,对IKONOS高空间分辨率图像进行分类,并与最大似然法和单源数据(光谱)SVM分类结果进行定性和定量比较分析。研究结果表明,多源数据复合的SVM高空间分辨率遥感图像分类方法,能够有效解决单源数据信息图像分类效果破碎的问题;总精度达到68.38%,Kappa达到0.5993;对高维输入向量具有高的推广能力;比单源信息的SVM和最大似然方法图像分类精度更高,适合高空间分辨率遥感图像分类。The RS image shows a very promising perspective for urban land-cover and land-use classification, particularly with very high resolution（1-4m） satellite images, while the traditional extraction methods of the high spatial resolution image has the shortcomings of the low accuracy and classification efficiency. This paper deals with the high spatial resolution image（IKONOS） classification based on the SVM method integrating the information of spectral, texture and structure. And comparing to the results based on Maximum Likelihood and SVM method with single-source data, this shows that the high spatial resolution RS image classification based on SVM Method with muhi-source data can solve the image classification fragmentation which is based on the single-source data, spectrum, and has the good generalization ability with the high dimension vector. It has more accuration than the maximum likelihood method and SVM based on the single source data, adapts to the high spatial resolution RS Image classification.

关 键 词：高空间分辨率 SVM 最优超平面 纹理 结构 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]