摘要
在矿山远程监控系统中,由于监控摄像头位置和环境等因素限制,往往会导致图像中存在各种噪声,如椒盐噪声、高斯噪声等,这些噪声会对图像质量产生严重影响,同时也会给后续的图像分析和处理带来很大困难。因此,如何准确地去除噪声,提高图像质量,一直是矿山远程监控系统中的重要问题。生成对抗学习是一种基于对抗生成网络(Generative Adversarial Networks,GAN)的图像处理技术,可以有效去除图像中的噪声。据此,提出了一种基于生成对抗残差学习的矿山远程监控图像去噪算法。该算法首先通过GAN生成器学习得到一组残差图像,然后通过残差学习方式将原始图像与残差图像相加得到去噪后的图像。同时,为提高算法的鲁棒性和适用性,还引入了噪声分布估计网络和自适应控制机制。试验结果表明:该算法可以有效去除矿山远程监控图像中的噪声,并且具有较好的鲁棒性和适用性。
In the remote monitoring system of mine,due to the restrictions of the location and environment of the monitoring camera,there are often various noises in the image,such as salt and pepper noise,Gaussian noise,etc.,which will have a serious impact on the image quality,but also bring great difficulties to the subsequent image analysis and processing.Therefore,how to accurately remove noise and improve image quality is always an important problem in mine remote monitoring system.Generative adversarial learning is an image processing technique based on Generative Adversarial Networks(GAN),which can effectively remove noise from images.In this paper,a mine remote monitoring image denoising algorithm based on generation residuals learning is proposed.The algorithm first learns a group of residual images by GAN generator,and then adds the original image and residual image by the way of residual learning to get the denoised image.At the same time,in order to improve the robustness and applicability of the algorithm,noise distribution estimation network and adaptive control mechanism are introduced.The experimental results show that the algorithm can effectively remove the noise in the remote monitoring image of mine,and has good robustness and applicability.
作者
樊培利
王建军
艾薇
FAN Peili;WANG Jianjun;AI Wei(Department of Information Engineering,Shanxi Water Conservancy Vocational and Technical College,Yuncheng 044000,China;School of Mechanical and Power Engineering,Henan Polytechnic University,Jiaozuo 454000,China;Shanxi Yellow River Water Affairs Bureau of the Yellow River Water Conservancy Commission,Yuncheng 044000,China)
出处
《金属矿山》
CAS
北大核心
2024年第5期286-292,共7页
Metal Mine
基金
教育部高校科技研究发展中心项目(编号:ZJXF2022256)。
关键词
矿山远程监控图像
生成对抗网络
残差学习
图像去噪
mine remote monitoring image
generative adversarial network
residual learning
image denoising
作者简介
樊培利(1983-),男,讲师,硕士;通信作者:王建军(1979-),男,讲师,硕士。
