| 基于邻域相似性的快速非局部滤波算法 |
| |
| 引用本文： | 许俊玲,罗晓军.基于邻域相似性的快速非局部滤波算法[J].商丘职业技术学院学报,2009,8(5):56-58. |
| |
| 作者姓名： | 许俊玲 罗晓军 |
| |
| 作者单位： | 1. 长沙理工大学数学与计算科学学院,湖南,长沙,410076
2. 长沙理工大学数学与计算科学学院,湖南,长沙,410076;商丘职业技术学院,河南,商丘,476000 |
| |
| 基金项目： | 河南省教育厅自然科学基金资助项目 |
| |
| 摘 要： | 在考虑像素邻域相似性的基础上,对Morel等人提出的非局部均值滤波算法进行优化.该算法充分考虑图像中存在的大量冗余信息,通过计算邻域内像素点灰度值的加权平均来恢复中心点的灰度值,并对邻域内的点进行一次预分类,只考虑邻域内与中心点相似性大的点,而忽略与中心点相似性小的点,从而减小了算法的计算复杂度.实验结果表明,用该算法处理的图像,其视觉效果和峰值信噪比均不低于非局部均值算法,而运行速度提高了45%以上.
|
| 关 键 词： | 数字图像 图像噪音 图像去噪 相似性 |
Fast Non--Local Mean Filter Based on the Neighbor Similarity Algorithm |
| |
| Institution: | XU Jun -ling, LUO Xiao -jun ( 1. School of Math. and Computing Science, Changsha University of Science and Technology, Changsha 410076, China ; 2. Shangqiu Vocational and Technical College, Shangqiu 476000, China) |
| |
| Abstract: | A fast Non - local Mean filter is proposed, this algorithm seriously consider the similarity of the pixels value in the neighbor region. which is the an efficient version of the Morel＇s Non - local Mean filter. This algorithm seriously consider the large number of the redundant information in one image, and compote the weighted average of the neighbor＇s pixels value, use which to recover the center pixel value. Before this, it makes a presert on the neighbor＇s pixel, ones are the small similarity pixels, the others are the big similarity pixels, only the big similarity pixels can be used to recover the center pixel, thus the algorithm＇s computational complexity is greatly reduced. The results show that this algorithm not only visual effect, but also PSNR is better than the original algorithm, and the computation time is raised above 45%. |
| |
| Keywords: | digital image image noise image denoising similarity |
| 本文献已被 维普 万方数据 等数据库收录！ |
|
