无人机航摄影像盲超分辨率重建技术研究  

Research on Blind Super-Resolution Reconstruction Technology for UAV Aerial Photography Image

作　　者：邢鹏威 邹婉棋 韦曾义 温成龙 XING Pengwei;ZOU Wanqi;WEI Zengyi;WEN Chenglong(College of Resources and Environment,Beibu Gulf University,Qinzhou 535011,Guangxi,China;Qinzhou Municipal Bureau of Natural Resources,Qinzhou 535000,Guangxi,China;China Coal Tianjin Design Engineering Co.,Ltd.,Tianjin 300100,China)

机构地区：[1]北部湾大学资源与环境学院,广西钦州535011 [2]钦州市自然资源局,广西钦州535000 [3]中煤天津设计工程有限责任公司,天津300100

出　　处：《工程技术研究》2024年第19期216-218,共3页Engineering and Technological Research

摘　　要：为解决在无人机倾斜摄影过程中航摄影像出现不同程度的退化,导致倾斜摄影测量精度较低的问题,文章提出对无人机原始影像进行模糊、加噪、下采样等退化处理的方法,利用盲超分辨率网络重建技术,进行影像重建,以提高影像的清晰度和分辨率,进而提高无人机倾斜摄影测量精度。对超分辨率重建影像进行空中三角测量解算和平差后,检查点的均方根误差(RMSE)最大降低了0.024 m,三维误差最大降低了0.021 m,水平误差最大降低了0.019 m,高程误差最大降低了0.022 m,这说明应用盲超分辨率重建技术对于提高无人机倾斜摄影测量数字成果的精度具有一定的实用价值。In order to solve the problem that the aerial photography image has diff erent degrees of degradation in the process of UAV oblique photography,which leads to the low accuracy of oblique photogrammetry,this paper proposes degradation processing methods such as blurring,noise addition,downsampling for the UAV original image.The Blind Super-Resolution network reconstruction technology is used to reconstruct the image,so as to improve its clarity and resolution,and then improve the accuracy of UAV oblique photogrammetry.After the aerial triangulation solution and adjustment is carried out for the Super-Resolution reconstructed image,the Root Mean Squared Error(RMSE)of the checkpoint is reduced by 0.024 m at most,the three-dimensional error is reduced by 0.021 m at most,the horizontal error is reduced by 0.019 m at most,and the elevation error is reduced by 0.022 m at most,which indicates that the application of Blind Super-Resolution reconstruction technology has certain practical value for improving the accuracy of UAV oblique photogrammetry numerical results.

关 键 词：倾斜摄影测量 空中三角测量 盲超分辨率重建 测量精度 

分 类 号：P231[天文地球—摄影测量与遥感]