计算机应用 ›› 2015, Vol. 35 ›› Issue (3): 835-839.DOI: 10.11772/j.issn.1001-9081.2015.03.835

• 虚拟现实与数字媒体 • 上一篇    下一篇

自适应加权编码L1/2正则化的图像重建算法

查志远, 刘辉, 尚振宏, 李润鑫   

  1. 昆明理工大学 信息工程与自动化学院, 昆明 650500
  • 收稿日期:2014-09-25 修回日期:2014-11-05 出版日期:2015-03-10 发布日期:2015-03-13
  • 通讯作者: 刘辉
  • 作者简介:查志远(1987-),男,江苏涟水人,硕士研究生,主要研究方向:图像处理、模式识别;刘辉(1969-),男,云南罗平人,教授,博士,主要研究方向:计算机视觉、模式识别;尚振宏(1975-),男,河南三门峡人,副教授,博士,CCF会员,主要研究方向:计算机视频、图像处理;李润鑫(1985-),男,云南昆明人,讲师,博士,CCF会员,主要研究方向:计算机图像、信号处理优化方法
  • 基金资助:

    国家自然科学基金资助项目(61462052);云南省自然科学基金资助项目(KKSY201403049);中国科学院太阳活动重点实验室项目(KLSA201310);昆明市科技局项目(08S100310)

Image restoration algorithm of adaptive weighted encoding and L1/2 regularization

ZHA Zhiyuan, LIU Hui, SHANG Zhenhong, LI Runxin   

  1. School of Information Engineering and Automation, Kunming University of Science and Technology, Kunming Yunnan 650500, China
  • Received:2014-09-25 Revised:2014-11-05 Online:2015-03-10 Published:2015-03-13

摘要:

针对图像重建过程中噪声去除问题,提出一种自适应加权编码L1/2正则化重建算法。首先,考虑到许多真实图像中不仅含有高斯噪声,而且含有拉普拉斯噪声,设计一种改进的L1-L2混合误差模型(IHEM)算法,该算法兼顾了L1范数与L2范数的各自优点;其次,由于迭代过程中噪声分布会发生改变,设计一种自适应隶属度算法,该算法可以减少迭代次数和运算时间;利用一种自适应加权编码方法,该方法可以有效地去除含有重尾分布特性的拉普拉斯噪声;另外,设计一种L1/2正则化算法,该算法可以得到较稀疏的解。实验结果表明,相比IHEM算法,自适应L1/2正则化图像重建算法的峰值信噪比(PSNR)平均提高了3.46 dB,结构相似度(SSIM)平均提高了0.02,对含有多种噪声的图像处理具有比较理想的效果。

关键词: L1-L2混合误差模型, 自适应隶属度, 加权编码, 稀疏解, L1/2正则化

Abstract:

Aiming at the denoising problem in image restoration, an adaptive weighted encoding and L1/2 regularization method was proposed. Firstly, for many real images which have not only Gaussian noise, but have Laplace noise, an Improved L1-L2 Hybrid Error Model (IHEM) method was proposed, which could have the advantages of both L1 norm and L2 norm. Secondly, considering noise distribution change in the iteration process, an adaptive membership degree method was proposed, which could reduce iteration number and computational cost. An adaptive weighted encoding method was applied, which had a perfect effect on solving the noise heavy tail distribution problem. In addition, L1/2 regularization method was proposed, which could get much sparse solution. The experimental results demonstrate that the proposed algorithm can lead to Peak Signal-to-Noise Ratio (PSNR) about 3.5 dB improvement and Structural SIMilarity (SSIM) about 0.02 improvement in average over the IHEM method, and it gets an ideal result to deal with the different noise.

Key words: L1-L2 Hybrid Error Model (HEM), adaptive membership degree, weighted encoding, sparse solution, L1/2 regularization

中图分类号: