一种新的图像复原方法

发布时间：2018-03-11 15:30

本文选题：图像复原　切入点：傅里叶变换　出处：《吉林大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着时代的不断发展,人们对图像质量的要求越来越高.图像复原技术作为改善图像质量的方法之一,已成为国内外研究的热点问题.图像的复原技术广泛地应用于天文、医学、军事、航空等领域,具有重大的现实意义.本文以传统图像复原算法为基础,探讨了几种经典的复原方法,提出了一种新的图像复原算法,复原图像的质量在数值实验和视觉效果上都有所提高.绪论中首先介绍了本文选题的研究意义,回顾了图像复原技术的研究背景、国内外的研究现状以及图像复原技术的最新进展,紧接着给出了本文的整体框架.第二章介绍了图像复原的基础知识、图像退化模型、退化函数估计、噪声类型以及评价指标.第三章介绍了几种经典的图像复原方法,比如维纳滤波、逆滤波、约束最小二乘法等.第四章是本文的核心内容,结合二对角变换以及寻找相似块的方法,提出一种新的图像复原方法.通过Householder变换二对角,使信号与噪声有效分离.第五章对本文提出的算法进行实验仿真.通过实验可以看出,本文提出的复原方法具有良好的效果,不仅保留了图像的主要特征信息,并且在一定程度上抑制了噪声.本文的最后一章对提出的复原算法进行评价总结并对未来研究方向和重点做了进一步的展望.
[Abstract]:With the development of the times, the demand for image quality is becoming higher and higher. As one of the methods to improve image quality, image restoration technology has become a hot issue in the research of domestic and foreign. Image restoration technology is widely used in astronomy. Medical, military, aviation and other fields have great practical significance. Based on the traditional image restoration algorithm, this paper discusses several classical restoration methods, and proposes a new image restoration algorithm. The quality of restored images has been improved in both numerical experiments and visual effects. In the introduction, the significance of this topic is introduced, and the research background of image restoration technology is reviewed. The research status at home and abroad and the latest development of image restoration technology are presented. Then the whole frame of this paper is given. In chapter 2, the basic knowledge of image restoration, image degradation model, degradation function estimation, and so on are introduced. Chapter 3 introduces several classical image restoration methods, such as Wiener filter, inverse filter, constrained least square method and so on. Chapter 4th is the core of this paper. A new image restoration method is proposed by combining the bidiagonal transformation and the method of finding similar blocks. In chapter 5th, the algorithm proposed in this paper is simulated. The experiments show that the restoration method proposed in this paper has a good effect, and not only retains the main feature information of the image, but also improves the performance of the image. In the last chapter of this paper, the proposed restoration algorithm is evaluated and summarized, and the future research directions and emphases are prospected.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41

【参考文献】