激光与光电子学进展, 2017, 54 (10): 101001, 网络出版: 2017-10-09   

基于深度学习的高光谱图像空-谱联合特征提取 下载: 1267次

Spatial-Spectral Feature Extraction of Hyperspectral Image Based on Deep Learning
作者单位
重庆大学光电技术及系统教育部重点实验室, 重庆 400044
摘要
由于高光谱遥感数据具有波段多、特征非线性、空间相关等特点, 提出一种基于深度学习的空-谱联合(SSDL)特征提取算法来有效提取数据中的空-谱特征。该算法利用多层深度学习模型--堆栈自动编码机对高光谱数据进行逐层学习, 挖掘图像中的深层非线性特征, 然后再根据每个特征像元的空间近邻信息, 对样本深度特征和空间信息进行空-谱联合, 增加同类数据聚集性和非同类数据分散度, 提升后续分类性能。在帕维亚大学和萨利纳斯山谷高光谱数据集上进行地物分类实验: 在1%样本比例下, 地物总体分类精度达到了91.05%和94.16%; 在5%样本比例下, 地物总体分类精度达到了97.38%和97. 50%。结果表明: 由于SSDL特征提取算法融合了数据中深层非线性特征和空间信息, 能够提取出更具鉴别特性的特征, 较其他同类算法能够获取更高分类精度。
Abstract
On the basis of the characteristics of multi-band, nonlinear and spatial correlation of hyperspectral remote sensing data, a new feature extraction algorithm based on spatial-spectral deep learning (SSDL) is proposed. This algorithm uses a multiple layers deep learning model, which is the stacked automatic encoder to study high spectral data layer by layer and explore the deep nonlinear characteristics of the image. Based on the spatial neighbor information of each feature pixel, the spatial-spectral combination of sample depth feature and spatial information is used to increase the compactness of homogeneous data and the separability of non-homogeneous data, and improve the performance of subsequent classification. The ground objects classification experiments are performed on Pavia University and Salinas Valley hyperspectral remote sensing datasets. When sample proportion is 1%, the ground objects overall classification accuracy reaches 91.05% and 94.16%. When sample proportion is 5%, the ground objects overall classification accuracy reaches 97.38% and 97.50%. The results show that the SSDL feature extraction algorithm fuses the deep nonlinear characteristics and spatial information of data. It can effectively extract the discriminant features, and obtain higher classification accuracy than other algorithms.
参考文献

[1] 孟鑫, 李建欣, 朱日宏, 等. 窄带高光谱干涉成像的压缩采样复原方法[J]. 光学学报, 2013, 33(1): 0130001.

    Meng Xin, Li Jianxin, Zhu Rihong, et al. Compressive sampling recovery method of narrow-band hyperspectral interferometric imaging[J]. Acta Optica Sinica, 2012, 33(1): 0130001.

[2] 赵春晖, 齐滨, 张燚. 基于改进型相关向量机的高光谱图像分类[J]. 光学学报, 2012, 32(8): 0828004.

    Zhao Chunhui, Qi Bin, Zhang Yi. Hyperspectral image classification based on variational relevance vector machine[J]. Acta Optica Sinica, 2012, 32(8): 0828004.

[3] Li T, Zhang J P, Zhang Y. Classification of hyperspectral image based on deep belief networks[C]. Proceedings of IEEE International Conference on Image Processing, Paris, 2014: 5132-5136.

[4] 樊利恒, 吕俊伟, 邓江生. 基于分类器集成的高光谱遥感图像分类方法[J]. 光学学报, 2014, 34(9): 0910002.

    Fan Liheng, Lü Junwei, Deng Jiangsheng. Classification of hyperspectral remote sensing images based on bands grouping and classification ensembles[J]. Acta Optica Sinica, 2014, 34(9): 0910002.

[5] 黄鸿, 曲焕鹏. 基于半监督稀疏鉴别嵌入的高光谱遥感影像分类[J]. 光学 精密工程, 2014, 22(2): 434-442.

    Huang Hong, Qu Huanpeng. Hyperspectral remote sensing image classification based on SSDE[J]. Optics and Precision Engineering, 2014, 22(2): 434-442.

[6] Tang Y Y, Yuan H L, Li L Q. Manifold-based sparse representation for hyperspectral image classification[J]. IEEE Transactions on Geoscience and Remote Sensing, 2014, 52(12): 7606-7618.

[7] Rumelhart D E, Hinton G E, Williams R J. Learning representations by back-propagating errors[J]. Nature,1986, 323(6088): 533-536.

[8] Hinton G E, Salakhutdinow R R. Reducing the dimensionality of data with neural networks[J]. Science, 2006, 313(5786): 504-207.

[9] le Roux N, Bengio Y. Deep belief networks are compact universal approximators[J]. Neural Computation, 2010, 22(8): 2192-2207.

[10] Fauvel M, Tarabalka Y, Benediktsson J A, et al. Advances in spectral-spatial classification of hyperspectral images[J]. Proceedings of the IEEE, 2013, 101(3): 652-675.

[11] Tan K, Hu J, Li J, et al. A novel semi-supervised hyperspectral image classification approach based on spatial neighborhood information and classifier combination[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2015, 105: 19-29.

[12] Mohan A, Sapiro G, Bosch E. Spatially coherent nonlinear dimensionality reduction and segmentation of hyperspectral images[J]. IEEE Geoscience & Remote Sensing Letters, 2007, 4(2): 206-210.

[13] 魏峰, 何明一, 梅少辉. 空间一致性邻域保留嵌入的高光谱数据特征提取[J]. 红外与激光工程, 2012, 41(5): 1249-1254.

    Wei Feng, He Mingyi, Mei Shaohui. Hyperspectral data feature extraction using spatial coherence based neighborhood preserving embedding[J]. Infrared and Laser Engineering, 2012, 41(5): 1249-1254.

[14] Lu S C, Liu H P, Li C W. Manifold regularized stacked autoencoder for feature learning[C]. Proceedings of IEEE International Conference on Systems Man and Cybernetics, 2015: 2950-2955.

[15] 胡昭华, 宋耀良. 基于Autoencoder网络的数据降维和重构[J]. 电子与信息学报, 2009, 31(5): 1189-1192.

    Hu Zhaohua, Song Yaoliang. Dimensionality reduction and reconstruction of data based on autoencoder network[J]. Journal of Electronics & Information Technology, 2009, 31(5): 1189-1192.

[16] 杨钊霞, 邹峥嵘, 陶超, 等. 空-谱信息与稀疏表示相结合的高光谱遥感影像分类[J]. 测绘学报, 2015, 44(7): 775-781.

    Yang Zhaoxia, Zou Zhengrong, Tao Chao, et al. Hyperspectral image classification based on the combination of spectral-spatial feature and sparse representation[J]. Acta Geodaetica et Cartographica Sinica, 2015, 44(7): 775-781.

[17] Fu W, Li S T, Fang L Y. Spectral-spatial hyperspectral image classification via super pixel merging and sparse representation[C]. Proceeding of IEEE Geoscience and Remote Sensing Symposium, 2015: 4971-4974.

[18] Yang S Y, Jin P L, Li B, et al. Semisupervised dual-geometric subspace projection for dimensionality reduction of hyperspectral image data[J]. IEEE Transactions on Geoscience and Remote Sensing, 2014, 52(6): 3587-3593.

黄鸿, 何凯, 郑新磊, 石光耀. 基于深度学习的高光谱图像空-谱联合特征提取[J]. 激光与光电子学进展, 2017, 54(10): 101001. Huang Hong, He Kai, Zheng Xinlei, Shi Guangyao. Spatial-Spectral Feature Extraction of Hyperspectral Image Based on Deep Learning[J]. Laser & Optoelectronics Progress, 2017, 54(10): 101001.

本文已被 13 篇论文引用
被引统计数据来源于中国光学期刊网
引用该论文: TXT   |   EndNote

相关论文

加载中...

关于本站 Cookie 的使用提示

中国光学期刊网使用基于 cookie 的技术来更好地为您提供各项服务,点击此处了解我们的隐私策略。 如您需继续使用本网站,请您授权我们使用本地 cookie 来保存部分信息。
全站搜索
您最值得信赖的光电行业旗舰网络服务平台!