摘要本文主要研究的是图像去噪的一种常用方法(非局部均值滤波算法)的改进。希 望可以在保持非局部均值滤波算法的良好效果的前提下,提高非局部均值滤波的时间 效率。
首先介绍了图像去噪的相关背景,以及常用方法,然后说明了非局部均值滤波算 法的主要思路和实现。之后阐述了新的算法,使用的方法是利用余弦积分图像化简代 替原来的权重函数,减少运算量,提高程序运行速度。采用对比实验比较新旧算法, 让原方法和新方法处理同样大小,相同噪声情况下时各自所花费的时间,以及去噪效 果。分别比较程序运行时间和峰值信噪比。实验的平台选择 matlab 语言实现。通过 对比实验,分析实验结果,得出最终结论。68466
新的算法可以在保持与原方法相近的去噪效果的同时,大幅度提高滤波效率。最 后针对实验数据,找出新算法的缺陷以及改进方向
毕业论文关键词 图像处理 图像去噪 非局部均值滤波 余弦积分
Title Fast Nonlocal Means Denoising Based on Cosine Integral Images and Applications
Abstract
This paper talks about the improvement of non-local means filter for image denoising. We hope that the new method can decrease the cost of time and maintain the high quality of denoising at the same time.
At first, related image denoising background is provided. Then, We illustrate the standard non-local means algorithm. Afterwards, the new method is introduced , which uses cosine integral image to simplify the weight function in order to reduce calculations. Several experiments are performed by programming with Matlab.
Finally, it comes out that the new method has the comparable effects on image denoising and is faster than the original one. By analyzing the statistics, I realize the new method has its own defects and it will be improved again.
Keywords Image processing; Image denoising; Non local means filter; Cosine integral image;
目 次
1 绪 论 1
2 图像去噪基础 3
2.1 图像处理与图像去噪 3
2.2 图像噪声 3
2.3 图像去噪的意义 4
2.4 图像去噪的历史及现状 4
3 图像去噪的经典算法 6
3.1 空域滤波 6
3.2 频域滤波 7
3.3 小结 10
4 标准 NLM 算法 11
4.1 NLM 算法概述 11
4.2 NLM 算法思路 11
4.3 算法实现 12