中国邮电高校学报(英文) ›› 2012, Vol. 19 ›› Issue (5): 115-123.doi: 10.1016/S1005-8885(11)60308-7

• Others • 上一篇    下一篇

JPEG Stream Soft-Decoding Technique Based on Autoregressive Modeling

牛毅1,石光明2,2,王晓甜1,王立志1   

  1. 1. 西安电子科技大学
  • 收稿日期:2012-04-18 修回日期:2012-07-03 出版日期:2012-10-31 发布日期:2012-10-08
  • 通讯作者: 牛毅 E-mail:niuyi@mail.xidian.edu.cn
  • 基金资助:

    自然科学基金

JPEG Stream Soft-Decoding Technique Based on Autoregressive Modeling

NIU Yi1, SHI Guang-ming1, WANG Xiao-tian1, WANG Li-zhi1, GAO Da-hua3   

  1. 1. School of Electronic Engineering, Xidian University, Xi’an 710071, China 2. ECE McMaster University, Hamilton L8S 4K1, Canada 3. School of Science, Air Force Engineering University, Xi’an 710051, China
  • Received:2012-04-18 Revised:2012-07-03 Online:2012-10-31 Published:2012-10-08
  • Contact: Yi Niu E-mail:niuyi@mail.xidian.edu.cn

摘要:

本文研究了一种新的JPEG码流软解码算法。采用2维分段自回归(PAR)模型对图像进行建模,根据DCT系数量化步长作为约束,将普通的JPEG解码过程转化为一个带约束的优化问题。PAR模型作为优化目标中的一个重要正则项,是该软解器的核心。考虑到PAR模型的参数只能从解码得到的有损图像中预测。 为了减少编码失真对模型参数预测精度的影响,我们提出一种双边带加权算法来对PAR模型参数进行自适应的总体最小二乘预测。大量的实验表明,该软解码器不但可以明显的提升JPEG图像的解码效果,并且在各个码率下都优于现有的针对JPEG解码图像的后处理方法。

关键词:

解块效应,自回归模型,待约束优化,总体最小二乘,双边带加权

Abstract:

This paper introduces a new model-based soft decoding technique to restore the widely used JPEG streams. The image is modeled as a 2D piecewise stationary autoregressive process, and the decoding task is formulated as an optimization problem with the constraint given by the quantization intervals which freely available at the decoder. The autoregressive model serves as an important regularization term of the objective function of the optimization. And the autoregressive model parameters are solved on the decoded image locally using a weighted total least square method, where a novel bilateral dualside weighting scheme is proposed to minimize the influence of the blocking artifact on the final estimation. Extensive experimental results suggest that the proposed algorithm systematically improves the quality of JPEG images and also outperforms existing JPEG postprocessing algorithms in a wide bit-rate range both in terms of PSNR and subjective quality

Key words:

Image deblocking, autoregressive modeling, constrained optimization, total least squares, bilateral weighting

中图分类号: