欢迎访问《图学学报》 分享到：
图学学报

图学学报

• 图像与视频处理 • 上一篇    下一篇

基于各向异性全变分的迭代滤波算法

  

  1. 1. 河南理工大学计算机科学与技术学院，河南 焦作 454000；
    2. 广东省数据科学工程技术研究中心，广东 广州 510631
  • 出版日期:2018-04-30 发布日期:2018-04-30
  • 基金资助:
    国家自然科学基金项目(U1404103)；河南省教育厅科学技术研究重点项目(14A520029，16A520053)；河南理工大学创新型科研团队项目
    (T2014-3)；河南理工大学博士基金项目(B2016-40)

An Iterative Image Filter Based on Anisotropic Total Variation

  1. 1. College of Computer Science and Technology, Henan Polytechnic University, Jiaozuo Henan 454000, China;
    2. Guangdong Engineering Research Center for Data Science, Guangzhou Guangdong 510631, China
  • Online:2018-04-30 Published:2018-04-30

PDF (PC)

205

摘要： 空间邻近度和像素值相似度的双边滤波(BF)器在滤波时，由于其值域滤波核系数的计
算易受到噪声的干扰，在噪声水平较大时，直接使用噪声图像来指导核函数权值计算的方案不可行。
为此，提出一种结合各向异性全变分和BF 的图像去噪算法，将各向异性全变分算法与BF 算法相结
合，首先利用各向异性全变分算法对噪声图像进行处理，得到一幅边缘结构信息较为丰富的结果图
像，接着将该结果图像作为BF 算法的引导图像来指导值域滤波核系数的计算，为保证算法的稳定
性，对上述过程进行迭代处理。此外，为提高各向异性全变分算法的计算效率，引入了Split Bregman
迭代算法进行加速处理。实验表明，该算法能在较好去噪的同时，保留较多的边缘结构信息。

关键词: 图像去噪, 双边滤波, 各向异性全变分算法, Split Bregman 迭代方法, 结构保持能力

Abstract: Spatial proximity and similarity of the pixel values of bilateral filter in the filter based on
the calculation of the range of filter kernel coefficient is susceptible to noise interference. When the
noise level is high, the direct use of noise image to guide the kernel weight computation program is
not feasible. Therefore, in this paper, the anisotropic total variation and bilateral filtering are
combined. Firstly, the image is processed by the anisotropic total variation model, and the result
image with rich edge structure information is obtained. Then the calculation results of image as a
guide bilateral filtering image to guide the range of filter kernel coefficient. In order to ensure the
stability of the algorithm, the above process is iterated. In addition, in order to improve the
computational efficiency of the anisotropic total variation model, the Split Bregman iterative
algorithm is introduced to accelerate the computation. The experimental results show that the
proposed algorithm can preserve more edge information while denoising.

Key words: image denoising, bilateral filter, anisotropic total variation, Split Bregman iterative method;
structure preserve capability