基于D-S证据理论加权的遥感图像组合分类  

Classification of A Weighted Combination of Remote Sensing Image Based on D-S Evidence Theory

作　　者：陈延辉[1] 韩志刚[2] 

机构地区：[1]平顶山学院资源与环境科学学院,平顶山467000 [2]河南大学中澳地理信息分析与应用研究所,开封475004

出　　处：《科学技术与工程》2013年第20期5970-5973,5977,共5页Science Technology and Engineering

基　　金：河南省科技计划发展计划(112102210332)资助

摘　　要：为了提高了遥感图像识别率,提出一种D-S证据理论加权的遥感图像组合分类模型。首先提取遥感图像的纹理和颜色特征,然后分别对纹理和颜色特征建立分类模型,并得到相应特征的分类率,最后把单一特征分类正确率输入到D-S证据理论对它们进行融合,得到它们的权值,根据权值得到遥感图像最终分类结果。仿真结果表明,本文模型不仅提高了遥感图像分类率,而且加快了遥感图像分类的速度,在地理信息系统具有一定的应用价值。In order to improve remote sensing image classification accuracy, a remote sensing image classification algorithm based on dempster-shafer theory and least square support vector machine is proposed. Firstly, texture features and color features were extracted from images, and these features are input to the least square support vec- tor machine to learn and get classification accuracy of single feature, and the classification accuracy of single feature is taken as evidence, and dempster-shafer theory is used to get the final classification result of remote sensing ima- ges. The simulation results show that DS-LSSVM not only has improved the image classification accuracy, but also accelerated the speed of remote sensing image classification, and it has certain application value in geographic information system.

关 键 词：遥感图像 纹理特征 颜色特征 最小二乘支持向量机 证据理论 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]