深度嵌套式Transformer网络的高光谱图像空谱解混方法  

Deep embedded Transformer network with spatial-spectral information for unmixing of hyperspectral remote sensing images

作　　者：游雪儿 苏远超 蒋梦莹 李朋飞 刘东升 白晋颖 You Xueer;Su Yuanchao;Jiang Mengying;Li Pengfei;Liu Dongsheng;Bai Jinying(College of Geomatics,Xi’an University of Science and Technology,Xi’an 710054,China;Faculty of Electronic and InformationEngineering,Xi’an Jiaotong University,Xi’an 710049,China;PIESAT Information Technology Co.,Ltd.,Beijing 100195,China)

机构地区：[1]西安科技大学测绘科学与技术学院,西安710054 [2]西安交通大学电子与信息学部,西安710049 [3]航天宏图信息技术股份有限公司,北京100195

出　　处：《中国图象图形学报》2024年第8期2220-2235,共16页Journal of Image and Graphics

基　　金：国家自然科学基金项目(42001319);教育部产学研协同育人项目(220802313200859);陕西省教育厅科研计划项目(21JK0762)。

摘　　要：目的基于深度学习的解混方法在信息挖掘和泛化性能上优于传统方法,但主要关注光谱信息,对空间信息的利用仍停留在滤波、卷积的表层处理。这使得构建解混网络时需要堆叠多层网络,易丢失部分图像信息,影响解混准确性。Transformer网络因其强大的特征表达能力广泛应用于高光谱图像处理,但将其直接应用于解混学习容易丢失图像局部细节。本文基于Transformer网络提出了改进方法。方法本文以TNT(Transformer in Transformer)构架为基础提出了一种深度嵌套式解混网络(deep embedded Transformer network,DETN),通过内外嵌入式策略实现编码器中局部与整体空间信息共享,不仅保留了高光谱图像的空间细节,而且在编码器中只涉及少量卷积运算,大幅度提升了学习效率。在解码器中,通过一次卷积运算来恢复数据结构以便生成端元与丰度,并在最后使用Softmax层来保障丰度的物理意义。结果最后,本文分别采用模拟数据集和真实高光谱数据集进行对比实验,在50dB模拟数据集中平均光谱角距离和均方根误差取得最优值,分别为0.0386和0.0045,在真实高光谱数据集Samson、Jasper Ridge中取得最优平均光谱角距离,分别为0.1194,0.1027。结论实验结果验证了DETN方法的有效性和优势,并且能为实现深度解混提供新的技术支撑和理论参考。Objective In hyperspectral remote sensing,mixed pixels often exist due to the complex surface of natural objects and the limitation of spatial resolution of instruments.Mixed pixels typically refer to the situation where a pixel in the hyperspectral images usually contains multiple spectral features,which hinders the application of hyperspectral images in various fields such as target detection,image classification,and environmental monitoring.Therefore,the decomposi⁃tion(unmixing)of mixed pixels is a main concern in the processing of hyperspectral remote sensing images.Spectral unmixing aims to overcome the limitations of image spatial resolution by extracting pure spectral signals(endmembers)rep⁃resenting each land cover class and their respective proportions(abundances)within each pixel.It is based on a spectral mixing model at the sub-pixel level.The rise of deep learning has brought many advanced modeling theories and architec⁃ture tools to the field of hyperspectral mixed pixel decomposition and has also spawned many deep learning-based unmixing methods.Although these methods have advantages over traditional methods in terms of information mining and generaliza⁃tion performance,deep networks often need to combine multiple layers of stacked network layers to achieve optimal learn⁃ing outcomes.Therefore,deep networks may cause damage to the internal structure of the data during the training process,which leads to the loss of important information in hyperspectral data and affects the accuracy of unmixing.In addition,most existing deep learning-based unmixing methods focus only on spectral information,but the exploit of spatial informa⁃tion is still limited to surface processing stages such as filtering and convolution.In recent years,autoencoder has been one of the research hotspots in the field of deep learning,and many variant networks based on autoencoder networks have emerged.Transformer is a novel deep learning network with an autoencoder-like structure.It has garnered considerable attention

关 键 词：遥感图像处理 高光谱遥感 混合像元分解 深度学习 Transformer网络 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]