SIA OpenIR  > 机器人学研究室
水下图像处理及双目视觉关键技术研究
Alternative TitleResearch on Key Technology of Underwater Image Processing and Binocular Vision
王国霖1,2
Department机器人学研究室
Thesis Advisor田建东
Keyword水下图像 水下成像模型 颜色校正 浑浊介质 双目视觉
Pages80页
Degree Discipline机械电子工程
Degree Name硕士
2020-05-26
Degree Grantor中国科学院沈阳自动化研究所
Place of Conferral沈阳
Abstract作为海洋探索和资源开采的重要工具,水下机器人的发展极大地促进了海洋研究的进程。和陆地不同,水下成像环境十分复杂,散射和吸收作用严重制约了水下机器视觉的发展。本文从成像模型和图像增强两个方向出发,开展水下图像处理和双目视觉关键技术的研究,对于提高水下机器人环境感知能力有着重要意义。本文研究的主要内容为:(1)针对防水罩折射问题,从空气中相机成像模型和水下双目相机成像模型两方面分析了水下双目相机的标定方法,并通过实验验证了高阶畸变补偿折射方法在水下图像的极线校正中的有效性。(2)针对现有水下成像模型无法有效地描述水体对光线的选择性吸收作用的问题,提出基于双透射率水下成像模型的图像复原方法。将原单透射率分为直接分量透射率和后向散射分量透射率,通过红通道先验和背景光恢复方法从单幅图像中获得模型参数,并通过实验证明了获得的双透射率满足水下物理成像规律。最后将获得的参数带入双透射率水下成像模型,获得复原结果。实验结果表明本文提出的复原方法能够有效解决水下图像的色偏问题。(3)针对现有方法的普适性不足和增强方法未考虑局部信息的问题,在增强方法中引入局部信息,提出基于白平衡、亮度调节和对比度拉伸的水下图像增强方法。首先利用带有局部信息的颜色补偿方法获取无色偏的水下图像。然后通过基于二维指数伽马函数的校正方法校正多尺度高斯函数提取的亮度图,并将校正后的亮度图代入带有饱和度限制的自适用对比度增强方法获得初步增强图像。最后通过深度相关的对比度拉伸方法获得清晰无色偏水下图像。对比结果表明本文增强方法能有效解决水下图像的色偏、轻度浑浊和重度浑浊问题,并能提高双目视觉中的特征点匹配数量。
Other AbstractAs an important tool for ocean exploration and resource extraction, the development of underwater robots has greatly promoted the process of ocean research. Different from land, underwater imaging environment is very complex, and the influence of scattering and absorption has seriously restricted the development of underwater machine vision. Starting from the two directions of imaging model and image enhancement, this paper conducts research on key technologies of underwater image processing and binocular vision, which is of great significance for improving the environment perception of underwater robots. The main contents of this paper are: (1) Aiming at the refraction problem of waterproof cover, the calibration method of underwater binocular camera was analyzed from two aspects: the air camera imaging model and the underwater binocular camera imaging model, and the validity of the high-order distortion compensation refraction method in the polar correction of underwater image was verified by experiments. (2) In view of the existing underwater imaging model cannot be described well selective absorption of light by water bodies, based on a double transmission underwater imaging model of image restoration method is proposed. The original single transmission can be divided into direct component transmission and backscatter component transmission, through the red channel prior and background light recovery method to obtain model parameters from a single image, and through the experiment proved that the double transmission meet law of underwater imaging physics. Finally, the parameters obtained are taken into the double transmission underwater imaging model to obtain the recovery results. Experimental results show that the proposed restoration algorithm can effectively solve the color cast problem of underwater images. (3) Aiming at the lack of universality of the existing algorithm and the problem that the enhancement algorithm does not consider the local information, the underwater image enhancement method based on white balance, brightness adjustment and contrast enhancement is proposed by introducing the local information into the enhancement algorithm. Firstly, the underwater image without color cast is obtained by using the color compensation method with local information. Then the brightness map extracted by the multi-scale Gaussian function is corrected by the correction method based on the 2D exponential gamma function, and the corrected brightness map is substituted into the adaptive contrast enhancement algorithm with saturation limitation to obtain the preliminary enhanced image. Finally, a clear underwater image without color deviation is obtained by the depth-dependent contrast enhancement method. The comparison results show that the proposed algorithm can effectively solve the problem of color cast, slight turbidity and severe turbidity in underwater images, and can improve the matching number of feature points in binocular vision.
Language中文
Contribution Rank1
Document Type学位论文
Identifierhttp://ir.sia.cn/handle/173321/27135
Collection机器人学研究室
Affiliation1.中国科学院沈阳自动化研究所;
2.中国科学院大学
Recommended Citation
GB/T 7714
王国霖. 水下图像处理及双目视觉关键技术研究[D]. 沈阳. 中国科学院沈阳自动化研究所,2020.
Files in This Item:
File Name/Size DocType Version Access License
水下图像处理及双目视觉关键技术研究.pd(4263KB)学位论文 开放获取CC BY-NC-SAApplication Full Text
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[王国霖]'s Articles
Baidu academic
Similar articles in Baidu academic
[王国霖]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[王国霖]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.