## Message Board

Volume 47 Issue 8
Aug.  2022
Turn off MathJax
Article Contents

GUO Congzhou, LI Ke, LI He, TONG Xiaochong, WANG Xiwen. Deep Convolution Neural Network Method for Remote Sensing Image Quality Level Classification[J]. Geomatics and Information Science of Wuhan University, 2022, 47(8): 1279-1286. doi: 10.13203/j.whugis20200292
 Citation: GUO Congzhou, LI Ke, LI He, TONG Xiaochong, WANG Xiwen. Deep Convolution Neural Network Method for Remote Sensing Image Quality Level Classification[J]. Geomatics and Information Science of Wuhan University, 2022, 47(8): 1279-1286.

# Deep Convolution Neural Network Method for Remote Sensing Image Quality Level Classification

##### doi: 10.13203/j.whugis20200292
Funds:

The National Natural Science Foundation of China 41671409

• Author Bio:

GUO Congzhou, PhD, associate professor, specializes in the theory and method of deep learning and remote sensing image processing. E-mail: czguo0618@sina.cn

• Corresponding author: LI Ke, PhD, lecturer. E-mail: like@lsec.cc.ac.cn
• Publish Date: 2022-08-05
•   Objectives  The application and development of remote sensing image requires higher and higher image quality. Different processing methods and parameters are often needed for different quality remote sensing images, which is not suitable for the intelligent demand. Through the classification of remote sensing image quality level, it can provide prior information for remote sensing image processing, evaluate the objective quality of remote sensing image, assessment the effect of sensor imaging. With the development and popularization of deep learning theory, it is possible to evaluate the quality of digital images by using deep convolution neural network.  Methods  We propose a classification model of quality classification for remote sensing images based on deep convolution neural network. It is established by improving the feature extraction network and classification design.After the quality classification pretreatment, the classical deep learning method is used to detect the target, and the detection accuracy is obviously improved, which can effectively solve the problem of unbalanced quality of the training set data.  Results  The experimental results show that this proposed method is better than the traditional method. The highest score value of accuracy, recall, precision and F1_score can reach 0.976, 0.972, 0.974 and 0.973 on the remote sensing image data set of Northwestern Polytechnic University.  Conclusions  The classification of remote sensing image quality by convolution neural network extends the application field of deep learning. It provides a new method for the quality evaluation of remote sensing image. The classical deep learning method is used to detect the target, and the detection accuracy is obviously improved though quality classification. It provides a way to solve the problem of imbalance in remote sensing image quality.
•  [1] 闫利，胡修兵，陈长军，等. 多模态图像配准的梯度一致性算子[J]. 武汉大学学报·信息科学版, 2013, 38(8): 969-972 http://ch.whu.edu.cn/article/id/2729 Yan Li, Hu Xiubing, Chen Changjun, et al. An Operator of Gradient Consistency for Multimodal Image Registration[J]. Geomatics and Information Science of Wuhan University, 2013, 38(8): 969-972 http://ch.whu.edu.cn/article/id/2729 [2] Gao X B. Design and Implementation of Marine Automatic Target Recognition System Based on Visible Remote Sensing Images[J]. Journal of Coastal Research, 2020, 115(sp 1): 277 [3] 李烨，许乾坤，李克东. 面向图像复原的残差密集生成对抗网络新方法[J]. 小型微型计算机系统, 2020, 41(4): 830-836 Li Ye, Xu Qiankun, Li Kedong. New Method of Residual Dense Generative Adversarial Networks for Image Restoration[J]. Journal of Chinese Computer Systems, 2020, 41(4): 830-836 [4] Ledig C, Theis L, Huszár F, et al. Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network[C]//IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, HI, USA, 2017 [5] Dai J, Li Y, He K, et al. R-FCN: Object Detection Via Region-Based Fully Convolutional Networks[C]//Conference on Neural Information Processing Systems, Barcelona, Spain, 2016 [6] Shelhamer E, Long J, Darrell T. Fully Convolutional Networks for Semantic Segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(4): 640-651 [7] Pulgar F J, Rivera A J, Charte F, et al. On the Impact of Imbalanced Data in Convolutional Neural Networks Performance[M]// Cham: Springer, 2017 [8] Guo H X, Li Y J, Shang J, et al. Learning from Class-Imbalanced Data: Review of Methods and Applications[J]. Expert Systems with Applications, 2017, 73: 220-239 [9] Shermeyer J, van Etten A. The Effects of Super-Resolution on Object Detection Performance in Satellite Imagery[C]//IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), Long Beach, CA, USA, 2019 [10] 闫利，胡晓斌. 利用Contourlet-SSIM视觉模型的IKONOS图像质量评价研究[J]. 武汉大学学报·信息科学版, 2014, 39(1): 12-16 http://ch.whu.edu.cn/article/id/2842 Yan Li, Hu Xiaobin. Image Quality Assessment of IKONOS Images Based on Contourlet-SⅡM Model[J]. Geomatics and Information Science of Wuhan University, 2014, 39(1): 12-16 http://ch.whu.edu.cn/article/id/2842 [11] 马旭东，闫利，曹纬，等. 一种新的利用梯度信息的图像质量评价模型[J]. 武汉大学学报·信息科学版, 2014, 39(12): 1412-1418 http://ch.whu.edu.cn/article/id/3133 Ma Xudong, Yan Li, Cao Wei, et al. A New Image Quality Assessment Model Based on the Gradient Information[J]. Geomatics and Information Science of Wuhan University, 2014, 39(12): 1412-1418 http://ch.whu.edu.cn/article/id/3133 [12] Redmon J, Divvala S, Girshick R, et al. You Only Look Once: Unified, Real-Time Object Detection[C]//IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, NV, USA, 2016 [13] Redmon J, Farhadi A. Yolov3: An Incremental Improvement[J]. arXiv, 2018, DOI:  1804.02767 [14] 巩丹超. 基于NⅡRS的高分辨率光学卫星影像质量评估技术[C]//第十八届十三省市光学学术会议，上海, 2010 Gong Danchao. High Resolution Optical Satellite Image Quality Assessment Technology via NⅡRS[C]//The 18th Optical Conference of 13 Provinces and Cities, Shanghai, China, 2010 [15] 袁媛. 基于深度卷积神经网络的图像质量评价方法研究[D]. 武汉: 武汉大学, 2017 Yuan Yuan. Research on Image Quality Assessment Method via Deep Convolution Neural Network (D]. Wuhan: Wuhan University, 2017 [16] 李真伟，崔国忠，郭从洲，等. 任意形状曲线刃边的点扩散函数估计方法[J]. 测绘学报, 2019, 48(3): 352-362 https://www.cnki.com.cn/Article/CJFDTOTAL-CHXB201903011.htm Li Zhenwei, Cui Guozhong, Guo Congzhou, et al. An Algorithm for the Estimation of Point Spread Function Based on Curve Edge of Arbitrary Shape[J]. Acta Geodaetica et Cartographica Sinica, 2019, 48(3): 352-362 https://www.cnki.com.cn/Article/CJFDTOTAL-CHXB201903011.htm [17] Li L, Pan J, Lai W S, et al. Learning a Discriminative Prior for Blind Image Deblurring[C]//IEEE Conference on Computer Vision and Pattern Recognition, Salt Lake City, UT, USA, 2018 [18] Goodfellow I, Bengio Y, Courville A. Deep Learning[M]. Cambridge: The MIT Press, 2016 [19] Lin M, Chen Q, Yan S. Network in Network[J]. arXiv, 2013, DOI:  1312.4400 [20] Ciberlin J, Grbic R, Teslic N, et al. Object Detectection and Object Tracking in Front of the Vehicle Using Front View Camera[C]//Zooming Innovation in Consumer Technologies Conference, Novi Sad, Serbia, 2019 [21] 赵文强，孙巍. 基于S4-YOLO的海上目标检测识别方法[J]. 光学与光电技术, 2020, 18(4): 38-46 https://www.cnki.com.cn/Article/CJFDTOTAL-GXGD202004006.htm Zhao Wenqiang, Sun Wei. Detection and Recognition Method of Marine Target Based on S4-YOLO[J]. Optics & Optoelectronic Technology, 2020, 18(4): 38-46 https://www.cnki.com.cn/Article/CJFDTOTAL-GXGD202004006.htm
###### 通讯作者: 陈斌, bchen63@163.com
• 1.

沈阳化工大学材料科学与工程学院 沈阳 110142

Figures(6)  / Tables(4)

## Deep Convolution Neural Network Method for Remote Sensing Image Quality Level Classification

##### doi: 10.13203/j.whugis20200292
###### 1. Department of Basic, Information Engineering University, Zhengzhou 450001, China2. School of Surveying and Mapping, Information Engineering University, Zhengzhou 450001, China
Funds:

The National Natural Science Foundation of China 41671409

• Author Bio:

• ###### Corresponding author:LI Ke, PhD, lecturer. E-mail: like@lsec.cc.ac.cn

Abstract:   Objectives  The application and development of remote sensing image requires higher and higher image quality. Different processing methods and parameters are often needed for different quality remote sensing images, which is not suitable for the intelligent demand. Through the classification of remote sensing image quality level, it can provide prior information for remote sensing image processing, evaluate the objective quality of remote sensing image, assessment the effect of sensor imaging. With the development and popularization of deep learning theory, it is possible to evaluate the quality of digital images by using deep convolution neural network.  Methods  We propose a classification model of quality classification for remote sensing images based on deep convolution neural network. It is established by improving the feature extraction network and classification design.After the quality classification pretreatment, the classical deep learning method is used to detect the target, and the detection accuracy is obviously improved, which can effectively solve the problem of unbalanced quality of the training set data.  Results  The experimental results show that this proposed method is better than the traditional method. The highest score value of accuracy, recall, precision and F1_score can reach 0.976, 0.972, 0.974 and 0.973 on the remote sensing image data set of Northwestern Polytechnic University.  Conclusions  The classification of remote sensing image quality by convolution neural network extends the application field of deep learning. It provides a new method for the quality evaluation of remote sensing image. The classical deep learning method is used to detect the target, and the detection accuracy is obviously improved though quality classification. It provides a way to solve the problem of imbalance in remote sensing image quality.

GUO Congzhou, LI Ke, LI He, TONG Xiaochong, WANG Xiwen. Deep Convolution Neural Network Method for Remote Sensing Image Quality Level Classification[J]. Geomatics and Information Science of Wuhan University, 2022, 47(8): 1279-1286. doi: 10.13203/j.whugis20200292
 Citation: GUO Congzhou, LI Ke, LI He, TONG Xiaochong, WANG Xiwen. Deep Convolution Neural Network Method for Remote Sensing Image Quality Level Classification[J]. Geomatics and Information Science of Wuhan University, 2022, 47(8): 1279-1286.
• 遥感图像处理的方法大多基于同一个图像质量等级，比如遥感图像配准[1]、地标检测、目标识别[2]等，一般都是将不同数据来源、不同空间分辨率、不同光谱分辨率的数据区分开，分别采用不同的技术手段进行处理，甚至有些方法仅限于某类分辨率数据。很多图像复原方法都是在已知该图像是退化的前提下进行复原处理[3], 使用时需要针对不同质量等级图像设计不同的方法或者调整相应参数才能获取相应的处理效果，与全自动化的智能时代技术需求不相符。对遥感图像进行质量等级分类可以为遥感图像理解提供重要的先验信息，也可以为传感器成像能力检验、遥感图像质量评价提供科学依据。

随着深度学习在数字图像处理领域的广泛应用，利用深度卷积神经网络（deep convolution neural network, DCNN）方法，对数字图像进行复原重建[4]、特征提取、目标检测[5]和语义分割[6]等处理已经成为学者研究的主流方向。基于DCNN的数字图像处理采用数据驱动的学习训练模式，处理精度和效果受训练集图像的质量和类别影响很大。由于成像环境的影响，遥感图像获取的图像数据集很难保证在同一个质量级别，这不仅会影响其他质量等级的遥感图像的处理效果，还会影响整体业务系统的处理效果。该类问题产生的根本原因是数据质量不平衡，也是深度学习算法关注的重要问题[7-8]。针对数据质量不平衡问题，现有的方法一般都是通过大量扩充图像数据集、增加训练时间和迭代次数进行处理，最终获取某一个处理效果的平均值，并不能提高整体效果。或者对遥感图像进行复原或超分辨重建，提升图像质量后再进行顶层处理，虽然有一定效果，但依然会受异质遥感图像数据量的影响[9]

本文在现有遥感图像客观质量评价[10-11]和质量等级分类[12-13]的基础上，利用深度学习的分类机能，构建了一种用于遥感图像质量级别分类的深度卷积神经网络模型。该模型对遥感图像进行多个质量等级的分类，更加细致和准确。

Reference (21)

/