基于关联规则遥感影像的分类研究——以渭库绿洲为例

发布时间：2018-12-10 20:32

【摘要】：对遥感影像的计算机图像处理与识别技术是现阶段遥感技术研究的热点内容,本文对新疆渭干库车河流域不同分辨率的全色影像与多光谱影像融合并构建多源空间数据库,再通过与面向对象分类法相结合,进行图像分割、特征提取后将结果取值输入Matlab软件,编制相应的算法程序,构建关联规则分类器,探索分类规则并进行遥感影像的分类.同样基于Matlab软件平台,根据最大似然分类算法对同一研究区影像进行分类,并对分类结果进行精度检验.结果表明:相较传统分类方法,关联规则方法拓宽了传统逻辑推理分类方法中获取信息的途径,使分类更加智能化,提高并增强了影像分类速度与算法可靠性,分类精度得到显著提高,Kapp系数为0.83.因此,关联规则方法更适用于影像分类,可以作为研究区有效信息提取的支持手段.
[Abstract]:The computer image processing and recognition technology of remote sensing image is a hot topic of remote sensing technology at present. In this paper, the panchromatic and multispectral images with different resolution in Weigan and Kuqa River basin in Xinjiang are fused and multi-source spatial database is constructed. After feature extraction, the result is input into Matlab software, the corresponding algorithm program is compiled, association rule classifier is constructed, classification rules are explored and the remote sensing image classification is carried out. Also based on the Matlab software platform, the image of the same research area is classified according to the maximum likelihood classification algorithm, and the accuracy of the classification results is verified. The results show that compared with the traditional classification method, the association rule method broadens the way of obtaining information in the traditional logical reasoning classification method, makes the classification more intelligent, improves and enhances the speed of image classification and the reliability of the algorithm. The classification accuracy was improved significantly, and the Kapp coefficient was 0.83. Therefore, the method of association rules is more suitable for image classification and can be used as an effective information extraction method in the research area.
【作者单位】：新疆大学资源与环境科学学院;新疆绿洲生态重点实验室;
【基金】：国家自然科学基金(U1303381,41261090,41161063) 高分辨率对地观测重大专项(民用部分)(95-Y40B02-9001-13/15) 新疆维吾尔自治区青年科技创新人才培养工程(2013711014) 教育部新世纪优秀人才支持计划(NCET-12-1075);教育部长江学者计划创新团队计划(IRT1180) 霍英东青年教师基金项目(121018) 新疆维吾尔自治区科技支疆项目(201591101) 新疆维吾尔自治区重点实验室专项基金(2014KL005)
【分类号】：TP751

【相似文献】