基于多尺度CNN模型的多时相PolSAR图像作物分类

Multi-scale CNN Model for Crop Classification with Multi-temporal Fully PolSAR Images

农作物分类是偏振合成孔径雷达(PolSAR)数据的重要应用之一。由于单时相PolSAR数据获取的信息有限,因此,采用多时相PolSAR数据,其含有农作物生长周期更丰富的特征信息。针对多时相PolSAR数据在极化特征分解时造成的“维数灾难”问题,提出了一种非负性约束稀疏自编码器(NC-SAE)的特征压缩方法,用于对分解后的特征数据进行压缩,以获得分类所需的有效特征。此外,构建了一种多尺度特征分类网络(MSFCN),该网络可以提高农作物的分类性能,且优于目前传统的卷积神经网络和支持向量机方法。通过使用欧空局提供的数据进行仿真实验,对分类结果进行性能评估,并与传统方法比较。实验结果表明:所提的方法具有很好的农业应用前景。

Crop classification is one of the most significant applications of polarimetric synthetic aperture radar(PolSAR)data. Owing to the limited information obtained by single-temporal PolSAR data,multi-temporal data are used in this paper to further provide ample information within various crop growing stages. However,the polarization scattering decomposition of multi-temporal PolSAR data easily causes“dimension disaster”. In view of this,a neural network of sparse auto-encoder with non-negativity constraints(NC-SAE)is proposed to compress the data,yielding efficient features for accurate classification. A novel classifier of multi-scale feature classification network(MSFCN)is constructed to improve the classification performance,which is proved to be superior to the popular classifiers of convolutional neural networks(CNN)and supper vector machine(SVM). The performances of the proposed method are evaluated and compared with the traditional methods by using the simulated Sentinel-1 data provided by European Space Agency(ESA). The classification results indicate that the proposed method has a good prospect for agricultural applications.