基于3D-CAE的高光谱图像端到端空谱联合压缩与重建  

End-to-End Joint Spectral-spatial Compression and Reconstruction of Hyperspectral Images Using a 3D Convolutional Auto-encoder

作　　者：刘佳 陈林伟 李大明 乐意 韩东 Liu Jia;Chen Linwei;Li Daming;Le Yi;Han Dong(The 28th Research Institute of China Electronics Technology Group Corporation,Nanjing 210007,China)

机构地区：[1]中国电子科技集团公司第28研究所,南京210007

出　　处：《信息化研究》2022年第4期42-50,共9页INFORMATIZATION RESEARCH

摘　　要：高光谱图像压缩模型对图像压缩后进行存储和传输,从而减小通信设备的压力。然而,目前绝大多数高光谱图像压缩模型只考虑压缩或者重建的优化,构建压缩和重建端到端优化的压缩模型对于模型性能的提升至关重要。文章提出了一种基于3D卷积自编码器的高光谱图像压缩模型实现高光谱图像端到端空谱联合压缩。实验结果表明,相比2D卷积自编码器,3D卷积自编码器能够将高光谱重建图像的光谱角映射、峰值信噪比和结构相似度分别改进33.1%、11.5%和2.2%。Hyper spectral images(HSIs) must be pre-processed by a compression model, which reduces the pressure of storing and transmitting the huge data in applications. Whereas most of the existing methods consider only the compression or reconstruction requirements, an end-to-end optimization would simultaneously improve the performance of both requirements. This paper proposes a three-dimensional convolutional auto-encoder(3D-CAE) that precisely achieves end-to-end joint spectral-spatial compression and reconstruction of HSIs. In an experimental evaluation, the proposed method improved the spectral angle mapper(SAM), peak signal-to-noise ratio(PSNR), and structural similarity index measurement(SSIM) of the reconstructed HSIs by 33.1%, 11.5%, and 2.2%, respectively, relative to competitive method 2D-CAE.

关 键 词：高光谱图像 3D卷积自编码器 端到端优化 空谱联合压缩 重建 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]