摘要
针对高光谱遥感分类研究中面临的数据处理效率低、分类结果精度不高等难题,本文在引入极限学习机(ELM)算法的基础上,顾及其在噪声影响下出现稳健性降低的现象,进一步将主成分分析(PCA)应用于ELM的前端,从而构建了一种基于PCA和ELM的联合算法。将该算法与目前常用的神经网络和支持向量机进行对比分析发现:PCA-ELM分类结果的精度最高,其数据处理效率也较高,该算法具有较强的稳健性和泛化能力,适用于高光谱遥感信息的高效提取。
To deal with the problem of low efficiency and low accuracy in the classification of hyperspectral remote sensing images,the Extreme Learning Machine（ELM）was applied in this research.Further,considering its lower robustness under the influence of noise,the Principal Components Analysis（PCA）was used in the pre-processing stage of hyperspectral remote sensing data.We therefore proposed a joint algorithm based on PCA and ELM.Its performance was then compared with two commonly-used classification algorithms（i.e.,Neural Network and Support Vector Machine）.Results indicate that the PCA-ELM joint algorithm has the higher accuracy and the higher efficiency.With the higher robustness and the better generalization ability,this proposed joint algorithm is suitable to extract useful informations from the hyperspectral remote sensing data.
作者
李静
吴孔江
LI Jing;WU Kongjiang(The First Surveying and Mapping Institute of Guizhou Province,Guiyang Guizhou 550025,Chin)
出处
《北京测绘》
2018年第7期794-799,共6页
Beijing Surveying and Mapping
关键词
高光谱
分类
极限学习机
主成分分析
遥感
hyperspectral image
classification
extreme learning machine
principal component analysis
remote sensing
