基于深度学习的遥感图像去模糊研究  

Research on deblurring of remote sensing images based on deep learning

作　　者：张文彪 蒋作[1] ZHANG Wen-biao;JIANG Zuo(School of Mathematics and Computer Science,Yunnan Minzu University,Kunming 650500,China)

机构地区：[1]云南民族大学数学与计算机科学学院,云南昆明650500

出　　处：《云南民族大学学报（自然科学版）》2023年第4期480-484,共5页Journal of Yunnan Minzu University:Natural Sciences Edition

基　　金：国家自然科学基金(61866040).

摘　　要：在遥感图像的获取过程中,因多种因素的影响,如卫星的不同拍摄角度、光学传感系统、大气等因素都会引起遥感图像质量发生退化现象.为了获取高质量的遥感图像,需对退化的遥感图像进行图像质量增强.图像质量增强往往要求在恢复图像时,图像的空间细节信息和高级上下文信息保持平衡,二者保持平衡有利于解决图像退化问题.但现有的遥感图像质量增强算法在保留空间细节信息及高级上下文信息时,并不能有效的让二者保持平衡.为了更好的解决遥感图像的退化问题,增强其图像质量,提出多阶段的遥感图像去模糊算法,并通过实验验证算法的有效性.In the process of acquiring remote sensing image,the quality of remote sensing image degrades due to the influence of many factors,such as different shooting angle of satellite,optical sensing system,atmosphere and so on.In order to obtain high quality remote sensing images,image quality enhancement is needed for degraded remote sensing images.Image quality enhancement usually requires a balance between spatial detail information and high-level context information in image restoration,which is beneficial to solve the problem of image degradation.However,the existing remote sensing image quality enhancement algorithms can not effectively keep the balance between spatial details and high-level context information.In order to better solve the degradation of remote sensing image and enhance its image quality,this paper proposes a multi-stage remote sensing image defuzzification algorithm,and verifies the effectiveness of the algorithm through experiments.

关 键 词：遥感图像 图像去模糊 深度学习 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]