基于深度卷积神经网络和多核学习的遥感图像分类方法

王鑫; 李可; 宁晨; 黄凤辰

doi:10.11999/JEIT180628

基于深度卷积神经网络和多核学习的遥感图像分类方法

doi: 10.11999/JEIT180628

王鑫^1, ,,
李可¹,
宁晨²,
黄凤辰¹

1.
河海大学计算机与信息学院南京 211100
2.
南京师范大学物理科学与技术学院南京 210000

基金项目: 教育部中央高校基本科研业务费专项资金(2019B15314)，国家自然科学基金(61603124)，江苏省“六大人才高峰”高层次人才项目(XYDXX-007)，江苏省“333高层次人才培养工程”，江苏政府留学奖学金项目

详细信息

作者简介:
王鑫：女，1981年生，副教授，研究方向为图像处理、模式识别、计算机视觉、机器学习

李可：女，1996年生，硕士生，研究方向为深度学习理论

宁晨：男，1978年生，讲师，研究方向为机器学习和模式识别

黄凤辰：男，1964年生，副教授，研究方向为图像处理和分析

通讯作者:
王鑫　wang_xin@hhu.edu.cn

中图分类号: TP751
计量
- 文章访问数: 4174
- HTML全文浏览量: 1408
- PDF下载量: 353
- 被引次数: 0
出版历程
- 收稿日期: 2018-06-27
- 修回日期: 2018-12-28
- 网络出版日期: 2019-01-03
- 刊出日期: 2019-05-01

Remote Sensing Image Classification Method Based on Deep Convolution Neural Network and Multi-kernel Learning

1.
College of Computer and Information, Hohai University, Nanjing 211100, China
2.
School of Physics and Technology, Nanjing Normal University, Nanjing 210000, China

Funds: Fundamental Research Funds for the Central Universities (2019B15314), The National Natural Science Foundation of China (61603124), Six Talents Peak Project of Jiangsu Province (XYDXX-007), 333 High-Level Talent Training Program of Jiangsu Province, Jiangsu Province Government Scholarship for Studying Abroad

摘要

摘要:
为解决传统遥感图像分类方法特征提取过程复杂、特征表现力不强等问题，该文提出一种基于深度卷积神经网络和多核学习的高分辨率遥感图像分类方法。首先基于深度卷积神经网络对遥感图像数据集进行训练，学习得到两个全连接层的输出将作为遥感图像的两种高层特征；然后采用多核学习理论训练适合这两种高层特征的核函数，并将它们映射到高维空间，实现两种高层特征在高维空间的自适应融合；最后在多核融合特征的基础上，设计一种基于多核学习-支持向量机的遥感图像分类器，对遥感图像进行精确分类。实验结果表明，与目前已有的基于深度学习的遥感图像分类方法相比，该算法在分类准确率、误分类率和Kappa系数等性能指标上均有所提升，在实验测试集上3个指标分别达到了96.43%, 3.57%和96.25%，取得了令人满意的结果。
- 高分辨率遥感图像 /
- 分类 /
- 卷积神经网络 /
- 多核学习
Abstract:
To solve the problems of complex feature extraction process and low characteristic expressiveness of traditional remote sensing image classification methods, a high resolution remote sensing image classification method based on deep convolution neural network and multi-kernel learning is proposed. Firstly, the deep convolution neural network is constructed to train the remote sensing image data set to learn the outputs of two fully connected layers, which are taken as two high-level features of remote sensing images. Then, the multi-kernel learning is used to train the kernel functions for these two high-level features, so that they can be mapped to the high dimensional space, where these two features are fused adaptively. Finally, with the combined features, a remote sensing image classifier based on Multi-Kernel Learning-Support Vector Machine (MKL-SVM) is designed for remote sensing image classification. Experimental results show that compared with the existing deep learning based remote sensing classification methods, the proposed algorithm achieves improved results in terms of classification accuracy, error, and Kappa coefficient. On the experimental test set, the above three indicators reach 96.43%, 3.57%, and 96.25% respectively, and satisfactory results are obtained.
- High resolution remote sensing image /
- Classification /
- Convolution neural network /
- Multi-Kernel Learning(MKL)

HTML全文

图 1 本文算法的框图

下载: 全尺寸图片幻灯片

图 2 搭建的7层卷积神经网络

下载: 全尺寸图片幻灯片

图 3 21类高分辨率遥感图像示例

下载: 全尺寸图片幻灯片

图 4 conv1中96个卷积核可视化结果

下载: 全尺寸图片幻灯片

图 5 各卷积层学习得到的特征图

下载: 全尺寸图片幻灯片

图 6 3种算法的分类混淆矩阵

下载: 全尺寸图片幻灯片

表 1 3种算法各分类性能指标值

指标算法1 算法2 本文算法

Ac 0.8808 0.9337 0.9643
Er 0.1192 0.0663 0.0357
Kappa系数 0.8748 0.9304 0.9625

下载: 导出CSV

参考文献(19)

WANG Xin, SHEN Siqiu, NING Chen, et al. Multi-class remote sensing objects recognition based on discriminative sparse representation[J]. Applied Optics, 2016, 55(6): 1381–1394. doi: 10.1364/ao.55.001381

WANG Xin, XIONG Xingnan, NING Chen, et al. Integration of heterogeneous features for remote sensing scene classification[J]. Journal of Applied Remote Sensing, 2018, 12(1): 15–23. doi: 10.1117/1.jrs.12.015023

NING Chen, LIU Wenbo, ZHANG Gong, et al. Enhanced synthetic aperture radar automatic target recognition method based on novel features[J]. Applied Optics, 2016, 55(31): 8893–8904. doi: 10.1364/ao.55.008893

冯子勇. 基于深度学习的图像特征学习和分类方法的研究及应用[D]. [博士论文], 华南理工大学, 2016.

FENG Ziyong. Research and application of image feature learning and classification methods based on deep learning[D]. [Ph.D. dissertation], South China University of Technology, 2016.

林妙真. 基于深度学习的人脸识别研究[D]. [硕士论文], 大连理工大学, 2013.

LIN Miaozhen. Research on face recognition based on deep learning[D]. [Master dissertation], Dalian University of Technology, 2013.

奚雪峰, 周国栋. 面向自然语言处理的深度学习研究[J]. 自动化学报, 2016, 42(10): 1445–1465. doi: 10.16383/j.aas.2016.c150682

XI Xuefeng and ZHOU Guodong. A survey on learning for natural language processing[J]. Acta Automatica Sinica, 2016, 42(10): 1445–1465. doi: 10.16383/j.aas.2016.c150682

张建华. 基于深度学习的语音识别应用研究[D]. [硕士论文], 北京邮电大学, 2015.

ZHANG Jianhua. Research into speech recognition application based on deep learning[D]. [Master dissertation], Beijing University of Posts and Telecommunications, 2015.

刘大伟, 韩玲, 韩晓勇. 基于深度学习的高分辨率遥感影像分类研究[J]. 光学学报, 2016, 36(4): 306–314. doi: 10.3788/AOS201636.0428001

LIU Dawei, HAN Ling, and HAN Xiaoyong. High spatial resolution remote sensing image classification based on deep learning[J]. Acta Optica Sinica, 2016, 36(4): 306–314. doi: 10.3788/AOS201636.0428001

曲景影, 孙显, 高鑫. 基于CNN模型的高分辨率遥感图像目标识别[J]. 国外电子测量技术, 2016, 35(8): 45–50. doi: 10.19652/j.cnki.femt.2016.08.011

QU Xianying, SUN Xian, and GAO Xin. Remote sensing image target recognition based on CNN[J]. Foreign Electronic Measurement Technology, 2016, 35(8): 45–50. doi: 10.19652/j.cnki.femt.2016.08.011

周敏, 史振威, 丁火平. 遥感图像飞机目标分类的卷积神经网络方法[J]. 中国图象图形学报, 2017, 22(5): 702–708. doi: 10.11834/jig.160595

ZHOU Min, SHI Zhenwei, and DING Huoping. Aircraft classification in remote-sensing images using convolutional neural networks[J]. Journal of Image and Graphics, 2017, 22(5): 702–708. doi: 10.11834/jig.160595

黄洁, 姜志国, 张浩鹏, 等. 基于卷积神经网络的遥感图像舰船目标检测[J]. 北京航空航天大学学报, 2017, 43(9): 1841–1848. doi: 10.13700/j.bh.1001-5965.2016.0755

HUANG Jie, JIANG Zhiguo, ZHANG Haopeng, et al. Ship object detection in remote sensing images using convolutional neural networks[J]. Journal of Beijing University of Aeronautics and Astronsutics, 2017, 43(9): 1841–1848. doi: 10.13700/j.bh.1001-5965.2016.0755

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. doi: 10.1109/5.726791

GAO Ligang, CHEN Paiyu, and YU Shimeng. Demonstration of convolution kernel operation on resistive crosspoint array[J]. IEEE Electron Device Letters, 2016, 7(7): 870–873. doi: 10.1109/LED.2016.2573140

GIRSHICK R, IANDOLA F, DARRELL T, et al. Deformable part models are convolutional neural networks[C]. Proceedings of the 2015 IEEE Conference on Computer Vision and Pattern Recognition. Boston, USA, 2015: 437–446. doi: 10.1109/CVPR.2015.7298641.

HINTON G E. How neural networks learn from experience[J]. Scientific American, 1992, 267(3): 145–151. doi: 10.1038/scientificamerican0992-144

BOUVRIE J. Notes on convolutional neural networks[R]. Massachusetts: Center for biological and computational learning, 2006: 38–44.

RUMELHAR T D E, HINTON G E, and WILLIAMS R J. Learning representations by back-propagating errors[J]. Nature, 1986, 323(6088): 533–536. doi: 10.1038/323533a0

KRIZHEVSKY A, SUTSKEVER I, and HINTON G E. ImageNet classification with deep convolutional neural networks[C]. Proceedings of Advances in Neural Information Processing Systems 25, Lake Tahoe, USA, 2012: 1097–1105. doi: 10.1145/3065386.

NING Chen, LIU Wenbo, and WANG Xin. Infrared object recognition based on monogenic features and multiple kernel learning[C]. The 2018 IEEE International Conference on Image, Vision and Computing. Chongqing, China, 2018: 204–208.

施引文献

资源附件(0)

访问统计