SIA OpenIR  > 光电信息技术研究室
单帧红外图像的低频非均匀性噪声校正算法研究
Alternative TitleResearch on Single Image Based Nonuniformity Correction Method in Infrared Camera
王帅
Department光电信息技术研究室
Thesis Advisor赵耀宏
Keyword单帧红外图像 非均匀性噪声 小波分解 红外成像
Pages55页
Degree Discipline计算机应用技术
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在红外成像系统中,低频非均匀性噪声会严重影响红外系统的成像效果,其中光学镜头和相机壳体辐射的非均匀性噪声表现为低频的暗圈或暗斑,这种噪声严重影响了红外成像系统的成像质量,需要采用非均匀性校正技术进行校正,然而传统基于标定的方法受时间等因素的影响,会使原有的校正参数发生变化,导致无法有效地去除低频非均匀性噪声,使得红外成像系统的成像质量下降。本文的研究内容主要是利用基于单帧红外图像的非均匀校正技术对红外图像进行校正,最大程度上还原红外图像的场景内容,提高图像质量,为之后其他处理做好基础工作。本文首先对低频非均匀性噪声的来源进行分析,然后介绍了几种国内外主流的基于单帧图像的低频非均匀性噪声校正算法,并指出这些算法的特点,根据这些算法的局限性,提出两种基于单帧图像的低频非均匀性校正算法,分别是基于改进的双边滤波的迭代最小二乘方法和基于小波滤波核和交替方向乘子方法的非均匀性校正算法。第一种方法先对含噪图像进行差分处理,然后利用优化的双边滤波算法抑制高频场景信息,然后利用迭代最小二乘法得到噪声模型的参数。第二种方法首先对含噪图像进行小波滤波来抑制高频信息,然后利用场景和噪声梯度信息的稀疏特性建立L1正则化能量泛函,最后利用交替方向乘子方法求解该最优化问题,得到模型参数,从而得到估计的低频非均匀性噪声图像。在将几种典型的低频非均匀性校正算法和本文提出的两种算法做对比,使用粗糙度和非均匀性值等评价指标,验证本文算法的有效性。
Other AbstractIn the infrared imaging system, the low-frequency nonuniformity noise will seriously affect the imaging effect of the infrared system, and the low-frequency dark circle or dark spot is the nonuniformity noise radiated by the optical lens and camera shell, which seriously affects the imaging quality of the infrared imaging system, so it needs to be corrected by the nonuniformity correction technology. However, due to the influence of time and other factors, the traditional calibration method will change the original correction parameters, which will lead to the inability to effectively remove the low-frequency nonuniformity noise, resulting in the degradation of the imaging quality of the infrared imaging system. The main research content of this paper is to use the nonuniformity correction technology based on single frame infrared image to correct the image of infrared imaging system, restore the scene content of the original infrared image to the greatest extent, improve the image quality, and do the basic work for other image processing steps later. This paper first analyzes the sources of low-frequency nonuniformity noise, then introduces several mainstream low-frequency nonuniformity noise correction algorithms based on single frame image at home and abroad, and points out the characteristics of these algorithms. According to the limitations of these algorithms, two nonuniformity correction algorithms based on single frame image are proposed, which make the iterative least square calculation based on improved bilateral filtering respectively Method and nonuniformity correction algorithm based on wavelet filter kernel and difference information. In the first method, the noisy image is first processed by difference, then the high frequency scene information is suppressed by the optimized bilateral filtering algorithm, and then the parameters of the noise model are obtained by the iterative least square method. In the second method, firstly, the noisy image is filtered by wavelet to suppress the high-frequency information, then the L1 regularization energy functional of non-uniform noise is established by using the sparse characteristics of scene and noise gradient information, and finally the surface parameters are obtained by using the alternating direction multiplier method to solve the optimization problem, thus the low-frequency non-uniform noise image is obtained.
Language中文
Contribution Rank1
Document Type学位论文
Identifierhttp://ir.sia.cn/handle/173321/27116
Collection光电信息技术研究室
Affiliation中国科学院沈阳自动化研究所
Recommended Citation
GB/T 7714
王帅. 单帧红外图像的低频非均匀性噪声校正算法研究[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
单帧红外图像的低频非均匀性噪声校正算法研(2805KB)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
[王帅]'s Articles
Baidu academic
Similar articles in Baidu academic
[王帅]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[王帅]'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.