多特征融合下的高光谱图像混合卷积分类  

Hyperspectral Image Hybrid Convolution Classification under Multi-Feature Fusion

作　　者：熊余[1,2,3] 单德明 姚玉 张宇 XIONG Yu;SHAN Deming;YAO Yu;ZHANG Yu(School of Communication and Information Engineering,Chongqing University of Posts and Telecommunications,Chongqing 400065,China;Key Laboratory of Optical Communication and Networks in Chongqing,Chongqing 400065,China;Key Laboratory of Ubiquitous Sensing and Networking in Chongqing,Chongqing 400065,China)

机构地区：[1]重庆邮电大学通信与信息工程学院,重庆400065 [2]重庆高校市级光通信与网络重点实验室,重庆400065 [3]泛在感知与互联重庆市重点实验室,重庆400065

出　　处：《红外技术》2022年第1期9-20,共12页Infrared Technology

基　　金：国家自然科学基金资助项目(61401052);国家留学基金委资助项目(201608500030);重庆市教委科学技术研究资助项目(KJ1400418,KJ1500445);重庆邮电大学博士启动基金资助项目(A2015-09)。

摘　　要：针对现有高光谱遥感图像卷积神经网络分类算法空谱特征利用率不足的问题,提出一种多特征融合下基于混合卷积胶囊网络的高光谱图像分类策略。首先,联合使用主成分分析和非负矩阵分解对高光谱数据集进行降维;然后,将降维所得主成分通过超像素分割和余弦聚类生成一个多维特征集;最后,将叠加后的特征集通过二维、三维多尺度混合卷积网络进行空谱特征提取,并使用胶囊网络对其进行分类。通过在不同高光谱数据集下的实验结果表明,在相同20维光谱维度下,所提策略相比于传统分类策略在总体精度、平均精度以及Kappa系数上均有明显提升。To address the problem of insufficient utilization of spatial-spectrum features in existing convolutional neural network classification algorithms for hyperspectral remote sensing images,we propose a hyperspectral image classification strategy based on a hybrid convolution capsule network under multi-feature fusion.First,a combination of principal component analysis and non-negative matrix decomposition is used to reduce the dimensionality of a hyperspectral dataset.Second,the principal components obtained through dimensionality reduction are used to generate a multidimensional feature set through super-pixel segmentation and cosine clustering.Finally,the superimposed feature set is used to extract spatial-spectrum features through a two-dimensional and three-dimensional multi-scale hybrid convolutional network,and a capsule network is used to classify them.We performed experiments on different hyperspectral datasets,and the results revealed that under the same 20-dimensional spectral setting,the proposed strategy significantly improves the overall accuracy,average accuracy,and Kappa coefficient compared to traditional classification strategies.

关 键 词：图像分类 高光谱图像 降维 超像素 混合卷积胶囊网络 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]