卷积神经网络在高分辨率影像分类中的应用

doi:10.19788/j.issn.2096-6369.190107

Abstract

Abstract:

[Objective] Convolutional neural networks were applied to the classification of high-resolution remote sensing images and compared with traditional classification methods to assess the classification accuracy and applicability of different classification methods for high-resolution remote image sensing. [Methods] The convolutional neural network classification method was compared with four commonly used traditional ENVI classification methods: maximum likelihood, parallelepiped, K-means mean clustering, and traditional neural network approaches. The confusion results were examined to evaluate the effectiveness of the different classification methods. [Results] Among the four traditional ENVI classification methods, the classification accuracy of the traditional neural network and maximum likelihood methods was relatively good, the classification accuracy of k-means mean clustering and parallelepiped methods was relatively poor, and the convolutional neural networks method achieved higher classification accuracy. [Conclusion] Convolutional neural networks can extract feature information in high-resolution remote sensing image classification, improving the classification accuracy of remote image sensing.

Key words: GF-1, Convolutional Neural Network, Remote Sensing, Deep Learning

CLC Number:

TP399

Li Xianjiang, Chen Youqi, Zou Jinqiu, Shi Shuqin, Guo Tao, Cai Weimin, Chen Hao. Application of Convolutional Neural Networks in High-Resolution Image Classification[J].Journal of Agricultural Big Data, 2019, 1(1): 67-77.

Figures/Tables 13

Fig.1

Fig.2

Fig.3

Table 1

Fig.4

Table 2

Fig.5

Fig.6

Table 3

Table 4

Table 5

Table 6

Table 7

References 22

[1]	Pacifici F, Chini M, Emery W J . A neural network approach using multi-scale textural metrics from very high-resolution panchromatic imagery for urban land-use classification.[J]. Remote Sensing of Environment, 2009,113(6):1276-1292. doi: 10.1016/j.rse.2009.02.014
[2]	Luo B, Zhang L . Robust autodual morphological profiles for the classification of high-resolution satellite images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2014,52(2):1451-1462. doi: 10.1109/TGRS.2013.2251468
[3]	王秋全, 李向, 王岭玲 . 基于不良地质识别的分类方法研究与分析[J]. 地质科技情报, 2014,33(06):203-208.
	Wang Q Q, Li X, Wang L L . Research and Analysis of Classification Methods Based on Bad Geological Identification[J]. Geological Science and Technology Information, 2014,33(06):203-208.
[4]	臧淑英, 张策, 张丽娟 , 等. 遗传算法优化的支持向量机湿地遥感分类——以洪河国家级自然保护区为例[J]. 地理科学, 2012,32(04):434-441.
	Zan S Y, Zhang C, Zhang L J , et al. Wetland Remote Sensing Classification Using Support Vector Machine Optimized With Genetic Algorithm:Algorithm: A Case Study in Honghe Nature National Reserve[J]. Scientia Geographica Sinica, 2012,32(04):434-441.
[5]	Krizhevsky A, Sutskever I, Hinton G E. ImageNet classification with deep convolutional neural networks [C]// International Conference on Neural Information Processing Systems. Curran Associates Inc. 2012: 1097-1105.
[6]	Fukushima K . Neocognitron: A self-organizing neural network model for a mechanism of pattern recognition unaffected by shift in position[J]. Biological Cybernetics, 1980,36(4):193-202.
[7]	Hu F, Xia G S, Hu J , et al. Transferring Deep Convolutional Neural Networks for the Scene Classification of High-Resolution Remote Sensing Imagery[J]. Remote Sensing, 2015,7(11):14680-14707.
[8]	Vakalopoulou M, Karantzalos K, Komodakis N, et al. Building detection in very high resolution multispectral data with deep learning features [C]. Geoscience and Remote Sensing Symposium. IEEE, 2015: 1873-1876.
[9]	Maggiori E, Tarabalka Y, Charpiat G , et al. Convolutional Neural Networks for Large-Scale Remote-Sensing Image Classification[J]. IEEE Transactions on Geoscience & Remote Sensing, 2016,55(2):645-657. doi: 10.1109/TGRS.2016.2612821
[10]	Bo Huang, Bei Zhao, Yimeng Song . Urban land-use mapping using a deep convolutional neural network with high spatial resolution multispectral remote sensing imagery [J] Remote Sensing of Environment,Volume 214, 2018,Pages 73-86.
[11]	Yu Liu, Xun Chen, Hu Peng, Zengfu Wang . Multi-focus image fusion with a deep convolutional neural network[J]. Information Fusion, 2017,36. doi: 10.1016/j.inffus.2016.12.001
[12]	Wang B, Sun Y, Xue B , et al. A Hybrid Differential Evolution Approach to Designing Deep Convolutional Neural Networks for Image Classification[J]. 2018.
[13]	Oquab M, Bottou L, Laptev I, et al. Learning and Transferring Mid-level Image Representations Using Convolutional Neural Networks [C]// Computer Vision and Pattern Recognition. IEEE, 2014: 1717-1724.
[14]	吕超 . 基于深度学习的情感分析方法研究[D]. 湖南大学, 2017.
	Lu C . Research on Sentiment Analysis Method Based on Deep Learning[D]. Hunan University, 2017.
[15]	Maxime Oquab, Leon Bottou, Ivan Laptev , et al. Learning and Transferring Mid-Level Image Representations using Convolutional Neural Networks[J]. 2014: 1717-1724. doi: 10.1109/CVPR.2014.222
[16]	Yu K, Salzmann M . Second-order Convolutional Neural Networks[J]. Clinical Immunology & Immunopathology, 2017,66(3):230-8. doi: 10.1006/clin.1993.1030
[17]	Defferrard M, Bresson X, Vandergheynst P . Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering[J]. 2016.
[18]	Schlemper J, Caballero J, Hajnal J V, et al. A Deep Cascade of Convolutional Neural Networks for MR Image Reconstruction [C]// International Conference on Information Processing in Medical Imaging. Springer, Cham, 2017: 647-658.
[19]	Levie R, Monti F, Bresson X , et al. CayleyNets: Graph Convolutional Neural Networks with Complex Rational Spectral Filters[J]. 2017.
[20]	余春艳, 吴明晖, 吴明 . 遥感图像识别中粗糙集理论与神经网络的结合[J]遥感学报, 2004(04):331-338.
	Yu C Y, Wu M H, Wu M . Combining Rough Set Theory with Neural Network for Remote Sensing Image[J]. Journal of Remote Sensing, 2004(04):331-338.
[21]	沈跃, 徐慧, 刘慧 , 等. 基于K-means和近邻回归算法的Kinect植株深度图像修复[J]. 农业工程学报, 2016,32(19):188-194.
	Shen Y, Xu H, Liu H , et al. Kinect scanning plant depth image restoration based on K-means and K-nearest neighbor algorithms[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2016,32(19):188-194.
[22]	李贤江, 石淑芹, 蔡为民 , 等. 基于CA-Markov模型的天津滨海新区土地利用变化模拟[J]. 广西师范大学学报(自然科学版), 2018,36(03):133-143.
	Li X J, Shi S Q, Cai W M , et al. Simulation of Land Use Change in Tianjin Binhai New Area Based on CA-Markov model[J]. Journal of Uuangxi Normal University (Natural Scicncc Edition), 2018,36(3):133-143.

名称		优缺点
最大似然法	假设每一波段的每一类统计结果都呈正态分布,计算给定像元属于某训练样本的相似度,像元将自动归入似然度最大的一类中。	总体精度较高,但在处理高分辨率影像时其计算时间较长。
平行六面体法	利用所选训练样区的亮度值,构建多维平行六面体的数据结构空间模型,并对遥感影像内的其它地物像元进行数据分析,如果其它地物像元的光谱值与先前训练样区的光谱值相似,则将其划归为一类。	可以充分利用像元的光谱特征,但其标准差阈值是由每个地类的均值决定的,分类精度易受干扰。
传统神经网络（Neural network）	由一系列相互链接的神经单元组成,每一个单元都有一定数量的实值输入,同时产生单一的实数值输出,通过训练样区的不断模拟和训练,对影像进行分类,其主要包括输入层、输出层和一个隐含层^[20]。	精度较高,但处理数据耗时较长。
K-Means均值聚类	K-Means均值聚类属于非监督分类,主要依靠遥感影像中不同地物类别的光谱信息来提取地物特征,然后通过算法对地物特征进行差异计算,从而达到对影像分类的效果,是较为传统的原型目标函数聚类方法^[21],其迭代规则的调整主要是通过函数求极值的方法实现的。	处理数据的时间较快,但分类精度较低,且容易受到人为因素的干扰。

	最大似然分类精度验证
	错分误差(%)	漏分误差(%)	制图精度(%)	用户精度(%)	总体分类精度(%)	Kappa系数
林地	1.46	2.1	97.9	98.54	97.8992	0.9708
耕地	5.38	4.16	95.84	94.62
建设用地	3.65	1.91	98.09	96.35
水域	0	1.09	98.91	100
未利用地	10.23	0	100	89.77

	平行六面体分类精度验证
	错分误差(%)	漏分误差(%)	制图精度(%)	用户精度(%)	总体分类精度(%)	Kappa系数
林地	12.86	3.59	96.41	87.14	83.6277	0.7728
耕地	36.74	23.88	76.12	63.26
建设用地	7.22	58.34	41.66	92.78
水域	0	6.91	93.09	100
未利用地	29.27	26.58	73.42	70.73

	k-means分类精度验证
	错分误差(%)	漏分误差(%)	制图精度(%)	用户精度(%)	总体分类精度(%)	Kappa系数
林地	27.29	32.32	67.68	72.71	69.98	0.5749
耕地	61.55	36.52	63.48	38.45
建设用地	100	100	0	0
水域	1.78	0.64	99.36	98.22
未利用地	0	100	0	0

	传统神经网络分类精度验证
	错分误差(%)	漏分误差(%)	制图精度(%)	用户精度(%)	总体分类精度(%)	Kappa系数
林地	0.41	2.82	97.18	99.59	98.6878	0.9817
耕地	4.6	1.2	98.8	95.4
建设用地	1.33	1.37	98.63	98.67
水域	0.06	0.31	99.69	99.94
未利用地	0	3.16	96.84	100

Application of Convolutional Neural Networks in High-Resolution Image Classification

RichHTML

PDF (PC)

Like

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 13

References 22

Related Articles 10

Metrics

Comments

Recommended 0

所分地类	卷积神经网络		土地利用现状		传统神经网络
所分地类	面积(km²)	比例(%)	面积(km²)	比例(%)	面积(km²)	比例(%)
林地	427.12	27.00	556.51	35.01	401.96	25.41
耕地	829.19	52.42	531.58	33.44	850.96	53.80
建设用地	240.67	15.22	322.60	20.30	229.02	14.48
水域	75.33	4.76	173.00	10.88	93.23	5.89
未利用地	9.50	0.60	5.73	0.36	6.60	0.42
总计	1581.81	100.00	1589.42	100.00	1581.78	100.00

所分地类	土地利用现状数据实际像元个数		卷积神经网络		传统神经网络
所分地类	土地利用现状数据实际像元个数		像元个数	空间像元误差	像元个数	空间像元误差
林地		20853	6197	0.703	46103	-1.211
耕地		29824	19538	0.345	39319	-0.318
建设用地		14427	6046	0.581	23356	-0.619
水域		14222	339	0.976	13653	0.040
未利用地		1161	137	0.882	3928	-2.383
总计		80487	32257	3.487	126359	4.535

[1]	ZHANG Zhaoxu, XIAO Yue, GOU Wentao, CUI Jin. Study on Drought Monitoring and Spatiotemporal Change in Henan Province Based on Sun/solar-induced Chlorophyll Fluorescence Remote Sensing [J]. Journal of Agricultural Big Data, 2023, 5(1): 76-86.
[2]	ZHANG Zhaoxu, CUI Jin, GOU Wentao, XIAO Yue. Drought Monitoring and Spatiotemporal Changes Analysis in North China Plain Based on Temperature Vegetation Dryness Index [J]. Journal of Agricultural Big Data, 2023, 5(1): 95-107.
[3]	Jia He, Laigang Wang, Yan Guo, Yan Zhang, Xiuzhong Yang, Ting Liu, Hongli Zhang. Estimating the Leaf Area Index of Maize based on Unmanned Aerial Vehicle Multispectral Remote Sensing [J]. Journal of Agricultural Big Data, 2021, 3(4): 20-28.
[4]	Ayitula Maimaitizunong, Shuai Yanju, Haodong Wei, Zhen He, Qinxi Xiao, Qiong Hu, Baodong Xu, Liangzhi You, Cougui Cao, Lin Ling. Evaluation of Green Development of Rice-Based Cropping Systems Using Remote Sensing Data and the DNDC Model: Case Study of Qianjiang City [J]. Journal of Agricultural Big Data, 2021, 3(3): 33-44.
[5]	Qiang Li, Maofang Gao, Ying Fang. Research on the Construction of the Agricultural Big Data Information Platform [J]. Journal of Agricultural Big Data, 2021, 3(2): 24-30.
[6]	Mingxu Zhang, Yuan Chen, Tuya Xilin, Ru Zhang, Yaqiong Bi, Chunhong Zhang, Taotao Wu, Minhui Li. Application and Prospects for Big Data of Traditional Chinese Medicine Resources [J]. Journal of Agricultural Big Data, 2021, 3(1): 14-24.
[7]	Laigang Guo Yan Wang Lijun He Jia Yang Xiuzhong Zhang Hongli Zheng Guoqing Wang. Developing a Combined Map of the Spatial Distribution of Autumn Crops in Henan Province Using Multi-source Remote Sensing [J]. Journal of Agricultural Big Data, 2020, 2(1): 53-59.
[8]	Xiaoman Li,Yang Zhang,Qian Xu,Nengfu Xie. Bibliometrics-Based Analysis of Advances in Plant Phenomics Research [J]. Journal of Agricultural Big Data, 2019, 1(2): 64-75.
[9]	Lei Wu,Xiaohe Liang,Jisiguleng Wu,Rui Wang. Method and Agricultural Empirical Study of Query Reformulation Based on Word Embedding [J]. Journal of Agricultural Big Data, 2019, 1(2): 114-120.
[10]	Lingxu Zhang,Rui Han,Wenming Li,Yinxue Shi,Chi Liu. A Survey of Big Data Deep Learning Systems and a Typical Agricultural Application [J]. Journal of Agricultural Big Data, 2019, 1(2): 88-104.