基于深度学习的高分辨率遥感影像分类研究

High Spatial Resolution Remote Sensing Image Classification Based on Deep Learning

针对高空间分辨率遥感影像的分类问题,提出了基于深度学习的分类方法。该方法通过非下采样轮廓波变换计算影像的纹理特征,利用深度学习的常用模型—深度信念网络(DBN)对高分辨率遥感影像进行了基于光谱-纹理特征的分类,并与基于单源光谱信息的DBN分类方法、支持向量机(SVM)分类方法、传统神经网络(NN)分类方法进行了比较分析。研究结果表明:相对于单源光谱信息,利用影像的光谱-纹理特征能够有效提高高分辨率遥感影像的分类精度;相对于SVM、NN等分类方法,DBN能够更加准确地挖掘高分辨率遥感影像的空间分布规律,提高分类的准确度。

A classification method based on deep learning is proposed for the classification of high spatial resolution remote sensing images. The texture features of the images are calculated through nonsubsampled contourlet transform, the deep learning common model-deep belief networks （DBN） are used to classify the high spatial resolution remote sensing images based on spectral and texture features. The proposed method is compared with the DBN classification method based on single spectral information, the support vector machine （SVM） method and the traditional neural network （NN） classification method. Experimental results show that comparing with the single spectral information, the use of spectral and texture information can effectively improve the classification accuracy of high spatial resolution remote sensing images, and comparing with methods of SVM and NN, the DBN method can accurately explore the distribution law of the high spatial resolution remote sensing images and improve the accuracy of classification.