基于近邻图正则化的高光谱图像特征选择方法  

Hyperspectral image feature selection method based on regularization by neighbor graph

作　　者：彭艳斌[1] 郑志军[1] 邱薇薇 李晓勇[1] 潘志刚[1] 金诚[2] PENG Yan-bin;ZHENG Zhi-jun;QIU Wei-wei;LI Xiao-yong;PAN Zhi-gang;JIN Cheng(School of Information and Electronic Engineering, Zhejiang University of Science and Technology , Hangzhou 310023,China;School of Computer Science and Technology .Zhejiang University, Hangzhou 310027,China)

机构地区：[1]浙江科技学院信息学院,浙江杭州310023 [2]浙江大学计算机学院,浙江杭州310027

出　　处：《光电子．激光》2019年第2期192-199,共8页Journal of Optoelectronics·Laser

基　　金：国家自然科学基金(61379074&61505176);浙江省自然科学基金(LQ13F020015)资助项目

摘　　要：高光谱图像分类是近年来的研究热点。其数据的高维性引发了"维度灾难"问题。数据降维成为解决问题的关键。针对高光谱数据有标记训练样本点匮乏的特点,提出用无监督的特征选择方法对高光谱数据进行降维。该方法能够同时保持原始高光谱数据的判别能力和局部几何结构。为了保持判别能力,用所选特征对原始高光谱数据进行重构,利用重构误差最小化将特征选择问题转化为优化问题。为了保持局部几何结构,建立近邻图,并将其转化为正则项加入目标函数中。通过迭代梯度下降方法解此优化问题,得出优选特征子集参与高光谱图像分类识别任务。在真实数据集上的实验表明,新方法能够提高分类识别的精度。Due to the characteristics of lacking training sample points in hyperspectral data,an unsupervised feature selection method to reduce dimension for hyperspectral data is represented.This method can keep the discriminant ability and local geometric structure of original data at the same time.In order to keep the discriminant ability,the method reconstructs the original data using the selected features.Through minimizing the reconstruction error,feature selection problem is transformed into an optimization problem.In order to keep the local geometric structure,neighbor graph is established and transformed into regular term,which is added into the objective function.Iterative gradient descent method is used to solve this optimization problem.The optimal feature subset is used in hyperspectral image classification task.Experiments on real data sets show that the new method can improve the accuracy of classification.

关 键 词：高光谱图像 分类 降维 特征选择 近邻图 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]