高光谱影像中蕴含了丰富的地物信息，具备“图谱合一”的特点，帮助人们不断认知地物属性的本质特征。高光谱影像地物分类旨在对高光谱影像覆盖的地物类型进行准确划分。丰富的光谱信息和空间信息提高了高光谱影像的应用潜力，但也存在波段数量大导致的数据冗余等问题。监督学习是一类常用的遥感影像分类方法，其良好分类结果需要依赖大量标注的训练样本。针对这些问题，本文将多尺度特征提取和基于模型的迁移学习引入高光谱影像分类中，系统阐述了多尺度特征提取和迁移学习的基本原理，设计并实现了相应算法，在四个公开的高光谱数据集上进行实验，有效提高了高光谱影像的分类精度，主要研究内容和成果如下：

针对高光谱影像场景中地物尺度差异较大、混合像元较多导致分类精度低的问题，在特征金字塔的基础上设计了一种多尺度特征神经网络，通过“自底向上”的过程对输入的高光谱数据进行降采样，输出包含深层次抽象语义信息的特征图，然后将最后一步输出的特征图进行上采样，得到高空间分辨率包含底层次语义信息的特征图，这一过程成为“自顶向下”。将这两个过程的特征图进行融合就得到了同时包含底层次语义信息和高层次语义信息的特征图，作为全连接层的输入进行分类。结果表明与单一尺度的特征提取方法相比，当高光谱影像地物场景复杂时，多尺度特征融合策略显著提高了高光谱影像的分类精度。

为解决高光谱影像分类中训练样本不足的问题，提出了一种基于深度模型迁移的高光谱影像分类方法。针对模型迁移中源域和目标域样本同为高光谱数据的迁移场景，首先将源域数据集划分为不同比例的训练样本，在源域模型上进行预训练，其次采用少量标记的目标域样本在模型上进行微调，提取目标域样本空间的专属特征，对高光谱影像中的空间特征和光谱特征进行融合。结果表明在目标域标记训练样本不足的情况下，基于同源样本进行迁移能够有效改善其分类结果。

针对自然图像和高光谱影像之间的迁移任务，充分挖掘现有的自然图像数据集在模型迁移中的潜力，设计了一种非对称迁移学习（ACTL，Asymmetric Convolutional Transfer Learning）模型，将深度神经网络中常用的卷积核替换为非对称卷积块，增加了卷积核中心区域的权重，提高了现有模型的特征提取能力，采用最大均值差异作为模型的损失函数以减小不同数据集之间的差异，有效提升了少标注训练样本时的分类效果，为异构数据集间的迁移学习提供新的思路。

Hyperspectral images contain rich information and have the feature of high spectral resolution and high spatial resolution, which helps people constantly recognize the essential characteristics of ground object attributes. As an important application in the field of hyperspectral remote sensing, the classification technology of ground objects in hyperspectral images aims to accurately classify the classes of ground objects covered by hyperspectral images. Although the rich spectral and spatial information in hyperspectral images brings rich information, there are still problems such as band redundancy and "dimension disaster" caused by the large amount of information. At the same time, the phenomenon of the same spectral for

different surface features and different spectra for the same surface feature in hyperspectral image classification also restrict the classification accuracy of hyperspectral images. As a common method of hyperspectral image classification, supervised classification methods’ performance rely on a large number of labeled hyperspectral training samples. To solve these problems, this article introduce the multi-scale feature extraction and transfer learning based on the model into the classification of hyperspectral image. Systematically introduces the basic principle of multi-scale feature extraction and transfer learning, corresponding algorithm was designed and implemented. Four open hyperspectral dataset were used to verify algorithm performance. Experiment results show that the classification accuracy of hyperspectral images significantly improved. The main aspects of the study and results achieved are as follows:

The scale of objects in hyperspectral images is quite different, lead to the problem of low classification accuracy of mixed pixels. In this article, a multi-scale feature of neural network based in feature pyramid is proposed. Through the "bottom-up" process, the input hyperspectral data is down-sampled, and the feature map containing deep abstract semantic information is output. The feature map output in the last step is up-sampled to obtain a feature map with high spatial resolution containing low-level semantic information which called "top-down". The feature maps of these two processes are combined to obtain a feature map that contains both low-level semantic information and high-level semantic information, which is used as the input of the fully connected layer for classification. The results show that multi-scale features extraction method can better extract features of different scales compared with the single-scale feature extraction method, under the situation of complex ground object scene, which remarkably improves the classification performance of HSIs.

In order to solve that it is difficult to obtain better classification results for hyperspectral image classification when the labeled training samples are insufficient, a hyperspectral image classification method based on deep model transfer is proposed. For which the source domain and target domain samples are both hyperspectral data in model based transfer learning, the source domain dataset is divided into samples of different proportions and then pre-trained on the deep neural network. Using a few number of labeled target domain samples to fine-tune the model and extract the specific feature in the target domain, and the spatial and spectral feature is combined in the hyperspectral image. In the case of insufficient labeled training data, transfer learning between heterogeneous data can significantly improve the classification results.

Aiming at the transfer task between natural images and hyperspectral images, and fully seeking the potential of existing natural image data sets in model transfer, an asymmetric convolutional transfer learning (ACTL) model is proposed with replace the convolution kernel commonly used in deep neural networks with an asymmetric convolution block, which increase the weight of the central area of the convolution kernel and improve the feature extraction ability of the existing model. Maximum mean difference is used as the loss function of the model to reduce the difference between different dataset. Experiment result shows that ACTL effectively improves the classification effect when the training samples are less labeled and provides new ideas for transfer learning between heterogeneous dataset.

[1]杜培军,夏俊士,薛朝辉,谭琨,苏红军,鲍蕊.高光谱遥感影像分类研究进展[J].遥感学报,2016,20(02):236-256.

[2]童庆禧,张兵,张立福.中国高光谱遥感的前沿进展[J].遥感学报,2016,20(05):689-707.

[3]张兵.高光谱图像处理与信息提取前沿[J].遥感学报,2016,20(05):1062-1090.

[4]Marcello J, Ibarrola-Ulzurrun E, Gonzalo-Martín C, et al. Assessment of hyperspectral sharpening methods for the monitoring of natural areas using multiplatform remote sensing imagery[J]. IEEE Transactions on Geoscience and Remote Sensing, 2019, 57(10): 8208-8222.

[5]Lees K J, Artz R R E, Khomik M, et al. Using spectral indices to estimate water content and GPP in Sphagnum moss and other peatland vegetation[J]. IEEE Transactions on Geoscience and Remote Sensing, 2020, 58(7): 4547-4557.

[6]Kim Y, Kim Y. Hyperspectral image classification based on spectral mixture analysis for crop type determination[C]//IGARSS 2018-2018 IEEE International Geoscience and Remote Sensing Symposium. IEEE, 2018: 5304-5307.

[7]Potgieter A B, Watson J, Eldridge M, et al. Determining crop growth dynamics in sorghum breeding trials through remote and proximal sensing technologies[C]//IGARSS 2018-2018 IEEE International Geoscience and Remote Sensing Symposium. IEEE, 2018: 8244-8247.

[8]张成业, 秦其明, 陈理,等. 高光谱遥感岩矿识别的研究进展[J]. 光学精密工程, 2015, 023(008):2407-2418.

[9]刘德长, 邱骏挺, 闫柏琨,等. 高光谱热红外遥感技术在地质找矿中的应用[J]. 地质论评, 2018, v.64(05):148-158.

[10]Tu B, Zhou C, He D, et al. Hyperspectral classification with noisy label detection via superpixel-to-pixel weighting distance[J]. IEEE Transactions on Geoscience and Remote Sensing, 2020, 58(6): 4116-4131.

[11]刘冰,余旭初,张鹏强,谭熊.联合空-谱信息的高光谱影像深度三维卷积网络分类[J].测绘学报,2019,48(01):53-63.

[12]Shimoni M, Haelterman R, Perneel C. Hypersectral imaging for military and security applications: Combining myriad processing and sensing techniques[J]. IEEE Geoscience and Remote Sensing Magazine, 2019, 7(2): 101-117.

[13]Zhang X, Wen G, Dai W. A tensor decomposition-based anomaly detection algorithm for hyperspectral image[J]. IEEE Transactions on Geoscience and Remote Sensing, 2016, 54(10): 5801-5820.

[14]王雪. 基于深度生成对抗模型的高光谱遥感影像分类研究[D].中国矿业大学,2019.

[15]Hughes G. On the mean accuracy of statistical pattern recognizers[J]. IEEE transactions on information theory, 1968, 14(1): 55-63.

[16]梁苗苗. 基于多尺度空谱融合网络的高光谱图像分类[D].西安电子科技大学,2018.

[17]王立伟. 深度迁移学习在高光谱图像分类中的应用[D].浙江工业大学,2019.

[18]王相海,王顺,谢释铖,李业涛,陶兢喆,宋传鸣.高光谱图像光谱维结构相关性及稀疏重建模型[J].中国科学:信息科学,2021,51(03):449-467.

[19]吴斯凡. 基于多尺度多场景迁移学习的高光谱图像分类算法研究[D].哈尔滨工业大学,2019.

[20]Cortes C, Vapnik V. Support-vector networks[J]. Machine learning, 1995, 20(3): 273-297.

[21]Dempster A P, Laird N M, Rubin D B. Maximum likelihood from incomplete data via the EM algorithm[J]. Journal of the Royal Statistical Society: Series B (Methodological), 1977, 39(1): 1-22.

[22]Guyon I, Elisseeff A. An introduction to variable and feature selection[J]. Journal of machine learning research, 2003, 3(Mar): 1157-1182.

[23]Li W, Prasad S, Tramel E W, et al. Decision fusion for hyperspectral image classification based on minimum-distance classifiers in thewavelet domain[C]//2014 IEEE China Summit & International Conference on Signal and Information Processing (ChinaSIP). IEEE, 2014: 162-165.

[24]Leo B, Jerome H F, Richard A O, et al. Classification and regression trees[J].European Journal of Operational Research,1984,19(1):144－160.

[25]MacQueen J. Some methods for classification and analysis of multivariate observations[C]//Proceedings of the fifth Berkeley symposium on mathematical statistics and probability. 1967, 1(14): 281-297.

[26]Xie X L, Beni G. A validity measure for fuzzy clustering[J]. IEEE Transactions on pattern analysis and machine intelligence, 1991, 13(8): 841-847.

[27]Dartnall H J A, Lythgoe J N. The spectral clustering of visual pigments[J]. Vision research, 1965, 5(4-5): 81-100.

[28]师芸,马东晖,吕杰,李杰,史经俭.基于流形光谱降维和深度学习的高光谱影像分类[J].农业工程学报,2020,36(06):151-160+323.

[29]Camps-Valls G, Bruzzone L. Kernel-based methods for hyperspectral image classification[J]. IEEE Transactions on Geoscience and Remote Sensing, 2005, 43(6): 1351-1362.

[30]Camps-Valls G, Gomez-Chova L, Muñoz-Marí J, et al. Composite kernels for hyperspectral image classification[J]. IEEE geoscience and remote sensing letters, 2006, 3(1): 93-97.

[31]Ratle F, Camps-Valls G, Weston J. Semisupervised neural networks for efficient hyperspectral image classification[J]. IEEE Transactions on Geoscience and Remote Sensing, 2010, 48(5): 2271-2282.

[32]Wang L, Niu S, Gao X, et al. Fast high-order sparse subspace clustering with cumulative MRF for hyperspectral images[J]. IEEE Geoscience and Remote Sensing Letters, 2020, 18(1): 152-156.

[33]Chirilă-Berbentea V C, Neagoe V E. Semi-Supervized Hyperspectral Image Classification using Virtually Labeled Training Data based on the Improved Gaussian Mixture Clustering[C]//2018 10th International Conference on Electronics, Computers and Artificial Intelligence (ECAI). IEEE, 2018: 1-4.

[34]Hubel D H, Wiesel T N. Receptive fields, binocular interaction and functional architecture in the cat's visual cortex[J]. The Journal of physiology, 1962, 160(1): 106-154.

[35]LeCun Y, Bottou L, Bengio Y, et al. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE, 1998, 86(11): 2278-2324.

[36]Hinton G E, Srivastava N, Krizhevsky A, et al. Improving neural networks by preventing co-adaptation of feature detectors[J]. arXiv preprint arXiv:1207.0580, 2012.

[37]Simonyan K, Zisserman A. Very deep convolutional networks for large-scale image recognition[J]. arXiv preprint arXiv:1409.1556, 2014.

[38]He K, Zhang X, Ren S, et al. Deep residual learning for image recognition[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2016: 770-778.

[39]Szegedy C, Liu W, Jia Y, et al. Going deeper with convolutions[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2015: 1-9.

[40]Dong Z, Wang M, Wang Y, et al. Object detection in high resolution remote sensing imagery based on convolutional neural networks with suitable object scale features[J]. IEEE Transactions on Geoscience and Remote Sensing, 2019, 58(3): 2104-2114.

[41]Jiang Y, Chen L, Zhang H, et al. Breast cancer histopathological image classification using convolutional neural networks with small SE-ResNet module[J]. PloS one, 2019, 14(3): e0214587.

[42]赵扬. 基于深度神经网络的高光谱影像地物分类技术研究[D].中国科学技术大学,2019.

[43]张博. 基于联合特征和深度学习的遥感影像分类[D].西安电子科技大学,2019.

[44]杨帆. 基于生成对抗网络的高光谱影像小样本分类技术研究[D].战略支援部队信息工程大学,2020.

[45]Mikolov T, Chen K, Corrado G, et al. Efficient estimation of word representations in vector space[J]. arXiv preprint arXiv:1301.3781, 2013.

[46]Chen Y, Jiang H, Li C, et al. Deep feature extraction and classification of hyperspectral images based on convolutional neural networks[J]. IEEE Transactions on Geoscience and Remote Sensing, 2016, 54(10): 6232-6251.

[47]Chen Y, Lin Z, Zhao X, et al. Deep learning-based classification of hyperspectral data[J]. IEEE Journal of Selected topics in applied earth observations and remote sensing, 2014, 7(6): 2094-2107.

[48]Chen Y, Zhao X, Jia X. Spectral–spatial classification of hyperspectral data based on deep belief network[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2015, 8(6): 2381-2392.

[49]Yu C, Zhao M, Song M, et al. Hyperspectral image classification method based on CNN architecture embedding with hashing semantic feature[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2019, 12(6): 1866-1881.

[50]Romero A, Gatta C, Camps-Valls G. Unsupervised deep feature extraction for remote sensing image classification[J]. IEEE Transactions on Geoscience and Remote Sensing, 2015, 54(3): 1349-1362.

[51]Li W, Wu G, Zhang F, et al. Hyperspectral image classification using deep pixel-pair features[J]. IEEE Transactions on Geoscience and Remote Sensing, 2016, 55(2): 844-853.

[52]Szegedy C, Vanhoucke V, Ioffe S, et al. Rethinking the inception architecture for computer vision[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2016: 2818-2826.

[53]张祥东,王腾军,朱劭俊,杨耘.基于扩张卷积注意力神经网络的高光谱图像分类[J].光学学报,2021:1-17.

[54]Lee H, Kwon H. Contextual deep CNN based hyperspectral classification[C]//2016 IEEE International Geoscience and Remote Sensing Symposium (IGARSS). IEEE, 2016: 3322-3325.

[55]Hang R, Liu Q, Hong D, et al. Cascaded recurrent neural networks for hyperspectral image classification[J]. IEEE Transactions on Geoscience and Remote Sensing, 2019, 57(8): 5384-5394.

[56]陈迪,郭锐,刘士彬,马勇,姚武韬,曹州.基于长短期记忆网络模型的中分辨率遥感影像变化检测研究[J].地理与地理信息科学,2019,35(06):25-30.

[57]孙晨昕. 基于三维卷积胶囊网络的高光谱图像分类方法研究[D].哈尔滨工业大学,2020.

[58]Goodfellow I, Bengio Y, Courville A, et al. Deep learning[M]. Cambridge: MIT press, 2016.

[59]Jun G, Ghosh J. An efficient active learning algorithm with knowledge transfer for hyperspectral data analysis[C]//IGARSS 2008-2008 IEEE International Geoscience and Remote Sensing Symposium. IEEE, 2008, 1: I-52-I-55.

[60]Pan S J, Yang Q. A survey on transfer learning[J]. IEEE Transactions on knowledge and data engineering, 2009, 22(10): 1345-1359.

[61]郑鑫,潘斌,张健.可变形网络与迁移学习相结合的电力塔遥感影像目标检测法[J].测绘学报,2020,49(08):1042-1050.

[62]Lin J, He C, Wang Z J, et al. Structure preserving transfer learning for unsupervised hyperspectral image classification[J]. IEEE Geoscience and Remote Sensing Letters, 2017, 14(10): 1656-1660.

[63]Persello C, Bruzzone L. Kernel-based domain-invariant feature selection in hyperspectral images for transfer learning[J]. IEEE transactions on geoscience and remote sensing, 2015, 54(5): 2615-2626.

[64]He X, Chen Y, Ghamisi P. Heterogeneous transfer learning for hyperspectral image classification based on convolutional neural network[J]. IEEE Transactions on Geoscience and Remote Sensing, 2019, 58(5): 3246-3263.

[65]Krizhevsky A, Sutskever I, Hinton G E. Imagenet classification with deep convolutional neural networks[J]. Advances in neural information processing systems, 2012, 25: 1097-1105.

[66]Li S, Song W, Fang L, et al. Deep learning for hyperspectral image classification: An overview[J]. IEEE Transactions on Geoscience and Remote Sensing, 2019, 57(9): 6690-6709.

[67]岑奕,张立福,张霞,王跃明,戚文超,汤森林,张鹏.雄安新区马蹄湾村航空高光谱遥感影像分类数据集[J].遥感学报,2020,24(11):1299-1306.

[68]丁国绅, 乔延利, 易维宁,等. 基于高光谱图像的改进SIFT特征提取与匹配[J]. 光学精密工程, 2020, 028(004):954-962.

[69]Lin T Y, Dollár P, Girshick R, et al. Feature pyramid networks for object detection[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2017: 2117-2125.

[70]Ding X, Guo Y, Ding G, et al. Acnet: Strengthening the kernel skeletons for powerful cnn via asymmetric convolution blocks[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision. 2019: 1911-1920.

[71]Ruder S. An overview of gradient descent optimization algorithms[J]. arXiv preprint arXiv:1609.04747, 2016.

[72]魏祥坡、余旭初、张鹏强、职露、杨帆. 联合局部二值模式的CNN高光谱图像分类[J]. 遥感学报, 2020, v.24(08):71-80.

[73]Rostami M, Kolouri S, Eaton E, et al. Sar image classification using few-shot cross-domain transfer learning[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops. 2019: 0-0.

[74]Pan S J, Tsang I W, Kwok J T, et al. Domain adaptation via transfer component analysis[J]. IEEE Transactions on Neural Networks, 2010, 22(2): 199-210.

[75]Yosinski J, Clune J, Bengio Y, et al. How transferable are features in deep neural networks?[J]. arXiv preprint arXiv:1411.1792, 2014.

[76]Deng J, Dong W, Socher R, et al. Imagenet: A large-scale hierarchical image database[C]//2009 IEEE conference on computer vision and pattern recognition. IEEE, 2009: 248-255.

[77]Russakovsky O, Deng J, Su H, et al. Imagenet large scale visual recognition challenge[J]. International journal of computer vision, 2015, 115(3): 211-252.

[78]田彦平, 陶超, 邹峥嵘,等. 主动学习与图的半监督相结合的高光谱影像分类[J]. 测绘学报, 2015(8):919-926.

[79]Shi Y, Ma D, Lv J, et al. ACTL: Asymmetric Convolutional Transfer Learning for Tree Species Identification Based on Deep Neural Network[J]. IEEE Access, 2021, 9: 13643-13654.

[80]Shin H C, Roth H R, Gao M, et al. Deep convolutional neural networks for computer-aided detection: CNN architectures, dataset characteristics and transfer learning[J]. IEEE transactions on medical imaging, 2016, 35(5): 1285-1298.

[81]Wang Q, Meng Z, Li X. Locality adaptive discriminant analysis for spectral–spatial classification of hyperspectral images[J]. IEEE Geoscience and Remote Sensing Letters, 2017, 14(11): 2077-2081.

[82]Deng C, Xue Y, Liu X, et al. Active transfer learning network: A unified deep joint spectral–spatial feature learning model for hyperspectral image classification[J]. IEEE Transactions on Geoscience and Remote Sensing, 2018, 57(3): 1741-1754.

[83]李冠东, 张春菊, 王铭恺,等. 卷积神经网络迁移的高分影像场景分类学习[J]. 测绘科学, 2019, 044(004):116-123,174.