通信学报 ›› 2019, Vol. 40 ›› Issue (1): 43-50.doi: 10.11959/j.issn.1000-436x.2019015

• 学术论文 • 上一篇    下一篇

基于结构相似性的非参数贝叶斯字典学习算法

董道广,芮国胜,田文飚,康健,刘歌   

  1. 海军航空大学信号与信息处理山东省重点实验室,山东 烟台 264001
  • 修回日期:2018-12-31 出版日期:2019-01-01 发布日期:2019-02-03
  • 作者简介:董道广(1990- ),男,山东济南人,海军航空大学博士生,主要研究方向为Bayesian统计学习、压缩感知和蒸发波导反演。|芮国胜(1968- ),男,山东烟台人,博士,海军航空大学教授、博士生导师,主要研究方向为混沌通信系统及现代滤波理论。|田文飚(1987- ),男,江西南昌人,博士,海军航空大学副教授,主要研究方向为压缩感知、蒸发波导反演。|康健(1971- ),女,黑龙江哈尔滨人,博士,海军航空大学副教授,主要研究方向为信号处理及现代滤波理论。|刘歌(1991- ),女,山东威海人,海军航空大学博士生,主要研究方向为压缩感知、蒸发波导反演。
  • 基金资助:
    国家自然科学基金资助项目(41606117);国家自然科学基金资助项目(41476089);国家自然科学基金资助项目(61671016)

Nonparametric Bayesian dictionary learning algorithm based on structural similarity

Daoguang DONG,Guosheng RUI,Wenbiao TIAN,Jian KANG,Ge LIU   

  1. Signal and Information Processing Key Laboratory in Shandong,Navy Aviation University,Yantai 264001,China
  • Revised:2018-12-31 Online:2019-01-01 Published:2019-02-03
  • Supported by:
    The National Natural Science Foundation of China(41606117);The National Natural Science Foundation of China(41476089);The National Natural Science Foundation of China(61671016)

摘要:

相较于传统综合字典学习方法,非参数贝叶斯方法具有显著优势,但其对图像结构全局相似性和变异性的表示能力仍有较大提升空间。针对这个问题,提出了一种基于结构相似性的非参数贝叶斯字典学习算法,该算法基于图像结构的全局相似性对图像进行聚类处理,并在图像的字典稀疏表示中引入块结构特性,提升了字典的结构表示能力。实验表明,所提算法在图像去噪和压缩感知方面的性能均优于目前主流的几种无监督字典学习算法。

关键词: 非参数贝叶斯, 字典学习, 结构相似性, 图像去噪, 压缩感知

Abstract:

Though nonparametric Bayesian methods possesses significant superiority with respect to traditional comprehensive dictionary learning methods,there is room for improvement of this method as it needs more consideration over the structural similarity and variability of images.To solve this problem,a nonparametric Bayesian dictionary learning algorithm based on structural similarity was proposed.The algorithm improved the structural representing ability of dictionaries by clustering images according to their non-local structural similarity and introducing block structure into sparse representing of images.Denoising and compressed sensing experiments showed that the proposed algorithm performs better than several current popular unsupervised dictionary learning algorithms.

Key words: nonparametric Bayesian, dictionary learning, structural similarity, denoising, compressed sensing

中图分类号: 

No Suggested Reading articles found!