利用EMAP与多项式网络的高光谱影像分类

高光谱影像地物分类已成为高光谱的重要应用之一，然而如何在小样本时取得优秀的分类结果已成为研究的难点与热点。最近几年，深度学习理论开始用于高光谱数据分析。本文提出了一种基于扩展多属性剖面（ex tended multi‐attribute profile ，EMAPs）和深度多项式网络（polynomial networks）的高光谱影像分类方法。首先，EMAPs通过一系列的属性滤波器提取影像多种结构特性的形态学纹理特征，并与影像光谱特征结合构成新的特征矢量。接着利用深度多项式网络对新特征矢量进行学习，构建多层次网络结构，在迭代的过程中逐层降低训练误差，实现优秀的分类结果。高光谱影像分类实验表明，所提方法性能优于多种分类方法。

Hyperspectral image classification using extended multi attribute profiles and polynomial networks

Hyperspectral image classification has become one of important hyperspectral applications ,but how to get good classification results in the case of small sample size is still an open issue and attracts many research attentions .In recent year ,deep learning has been used in the context of remote sensing image analysis .In this paper ,we propose a new hyperspectral image classification method based on EMAPs (extended multi‐attribute profile) and polynomial networks (PN) .Firstly ,EMAPs can extract multi‐level structures of morphological features by a series of attribute filters ,which integrate the spatial and spectral information of remote sensing data .Then ,the spatio‐spectral features are fed as the input to the deep PN composed of multi‐layer feed forward structure .PN decreases the training error layer‐by‐layer in the sense to obtain good classification results . Classification experimental results on different hyperspectral image sets demonstrate that the proposed method outperforms other methods .