Hyperspectral image classification algorithm based on three-dimensional wavelet transform and random patches network
党琦 1刘德山 1闫德勤 1张宇1
扫码查看
点击上方二维码区域，可以放大扫码查看
作者信息
1. 辽宁师范大学计算机与信息技术学院,辽宁 大连 116029
折叠
摘要
针对如何充分利用空间特征来达到较高的高光谱图像分类精度的问题,提出了一种基于三维离散小波变换(3D-DWT)与随机补丁网络(RPNet)结合的高光谱图像的地物属性分类算法.在分类过程中,综合3D-DWT提取的特征和RPNet深度学习框架提取的特征,利用支持向量机(SVM)对特征向量进行分类.所提出的方法在Indian Pines和University of Pavia两个数据集上进行测试,结果表明该方法比现有方法有显著的分类性能的提高.
Abstract
Aiming at the problem of how to make full use of spatial features to achieve high precision,a hyperspectral image classification algorithm based on 3D discrete wavelet transform(3D-DWT)and random patches network(RPNet)was proposed.During the classification process,the features extracted from 3D-DWT and RPNet deep learning framework were integrated,and support vector machines(SVM)were used to classify the feature vectors.The method proposed was tested on Indian Pines and University of Pavia datasets.The results showed that the method has significantly better performance than the existing methods.
关键词
三维离散小波变换(3D-DWT)/随机补丁网络(RPNet)/支持向量机(SVM)/高光谱图像分类
Key words
3D discrete wavelet transform/random patches network/support vector machine/hyper-spectral image classification
维普
万方数据