Computer Engineering and Applications ›› 2020, Vol. 56 ›› Issue (14): 226-230.DOI: 10.3778/j.issn.1002-8331.2004-0042

Previous Articles     Next Articles

Research on Superposition Iterative Denoising Algorithm Based on Total Variation

YAN Ding, LV Donghao, ZHANG Yong   

  1. College of Information Engineering, Inner Mongolia University of Science and Technology, Baotou, Inner Mongolia 014010, China
  • Online:2020-07-15 Published:2020-07-14

全变分的叠加迭代降噪算法的研究

颜鼎,吕东澔,张勇   

  1. 内蒙古科技大学 信息工程学院,内蒙古 包头 014010

Abstract:

With respect to denoising of the sparse signal with large peak-to-peak transitions, this paper puts forward a denoising algorithm of superposition iteration after total variation. Such denoising algorithm can not only effectively mitigate noise of the signal in the large transition zone, but also retain its original frame. Specifically, the signal after continuous total variation-based denoising is differenced, and the resultant difference is superimposed based on the original signal, so that it not only covers the information of the original signal but also retains the previous denoising information. Furthermore, noise of the sparse signal with large peak-to-peak transitions will be effectively mitigated through iterative denoising. Compared with the optimization-minimum algorithm, the experimental results are closer to the original signal.

Key words: majorization-minimum algorithm, total variation denoising, differenced, superposition iteration

摘要:

针对峰峰值大范围跳变的稀疏信号的降噪问题,提出了一种全变分后叠加迭代的降噪算法。该算法能够有效地对大范围跳变区间的信号降噪,并保留信号的原始框架。具体而言,进行连续的全变分降噪,对连续降噪后的信号做差,在原信号的基础上对求得的差值进行叠加处理,使其既涵盖了原信号的信息,又保留了之前降噪后的信息。在此基础上再进行迭代,实现对峰峰值大范围跳变的稀疏信号的有效降噪。实验结果与优化最小值算法相比更加接近原始信号。

关键词: 优化-最小值算法, 全变分, 差值, 叠加迭代