摘要
针对航测作业过程中遇到云雾等恶劣气候条件时,会产生影像色彩失真、饱和度低、背景模糊不清等问题,本文进行了影像去雾与影像增强算法的研究,提出了一种改进Retinex算法,并以重庆某山区为例,与两种传统经典算法进行比较分析。结果表明,本文算法可对带雾影像进行有效处理,获得更真实准确的细节、更清晰真实的光感色彩;使用去雾与增强后的影像进行三维建模,从而生成高分辨率三维网格。分别从主观和客观开展对影像效果的评价,所得到的优化影像与三维模型都有良好的效果。
At present,in the southwestern mountainous regions of China,the terrain environment is quite intricate,and there is frequent rain and fog weather throughout the year.The presence of smog significantly impacts aerial images,resulting in darker images,reduced image contrast,and loss of valuable information in the details,leading to impaired image.quality and 3D reconstruction models.This study focuses on addressing challenging weather conditions during aerial survey operations,such as clouds and fog,which often cause color distortion,low saturation,and blurred background in images.To tackle these issues,the study investigates image defogging and enhancement algorithms,presenting an improved Retinex algorithm.The model-based image restoration and defogging method analyzes the principle of blurring images and the mechanism behind the degradation of image quality due to rain and fog,subsequently processing foggy images.This paper firstly analyzes and introduces the traditional classic defogging enhancement algorithms,including dark channel prior algorithm,histogram equalization algorithm and Retinex algorithm.It explores model-based image restoration and defogging methods,and examines the principle of blurred images and the quality of rain and fog images mechanism of degradation,focusing on processing foggy images.Furthermore,non-model-based image enhancement algorithm are utilized to suppress irrelevant features,enhance image clarity and contrast,and enrich color and other field information,achieving a visually satisfactory visual enhancement effect.After conducting on-site image data collection in a foggy area in Chongqing,experiments and result analyses are performed to compare image defogging enhancement,in three different application scenarios:urban scene,slope scene,and forest scene.Both subjective and objective comparisons are made,with objective measurements utilizing information entropy and average gradient of the images to reflect the defogging enhancement effect through experimental result data.The use of the improved Retinex algorithm in this paper for aerial image defogging and enhancement processing better meet the specific scene requirements in various different areas.The proposed algorithm in this paper effectively processes foggy images,revealing more real and accurate details,and preserving clear and authentic light-sensitive colors.The defogged and enhanced images are further employed for 3D modeling,generating high-resolution 3D meshes.The evaluation of the image effect is conducted based on both subjective and objective standards,yielding positive results in optimized images and 3D models.The method presented in this paper assists in increasing the production of UAV aerial survey operations and real-world 3D modeling,providing significant support for actual surveying,mapping operations,and scientific research.It offers important data support for the next step of real-scene 3D modeling,further promoting the comprehensive application of real-scene 3D China.
作者
龚洲
潘国兵
敖其勇
熊延
陈昌文
袁小彬
GONG Zhou;PAN Guobing;AO Qiyong;XIONG Yan;CHEN Changwen;YUAN Xiaobin(School of Smart City,Chongqing Jiaotong University,Chongqing 400000,China)
出处
《时空信息学报》
2023年第2期202-208,共7页
JOURNAL OF SPATIO-TEMPORAL INFORMATION
基金
国家自然科学基金面上项目(42074004)。
