电子学报 ›› 2012, Vol. 40 ›› Issue (9): 1795-1800.DOI: 10.3969/j.issn.0372-2112.2012.09.015

• 学术论文 • 上一篇    下一篇

基于局部自回归模型的压缩感知视频图像递归重建算法

李星秀1, 韦志辉2   

  1. 1. 南京理工大学理学院, 江苏南京 210094;
    2. 南京理工大学计算机学院, 江苏南京 210094
  • 收稿日期:2011-11-28 修回日期:2012-05-09 出版日期:2012-09-25
    • 作者简介:
    • 李星秀 女,1981年12出生于山西省大同市,现为南京理工大学理学院讲师.主要研究方向为稀疏表示、压缩感知. E-mail:xxlwpl@126.com 韦志辉 男,1963年11月出生于江苏省淮安市,现为南京理工大学计算机科学与技术学院教授、博士生导师.主要从事计算机图像处理、模式识别、几何多尺度分析等领域的研究工作. E-mail:gswei@njust.edu.cn
    • 基金资助:
    • 国家自然科学基金 (No.61071146,No.61101194,No.61101198); 江苏省自然科学基金 (No.BK2011701); 高等学校博士学科点专项科研基金 (No.20113219120024)

Compressed Sensing Video Images Recursive Reconstruction Algorithm Based on Local Autoregressive Model

LI Xing-xiu1, WEI Zhi-hui2   

  1. 1. School of Science, Nanjing University of Science and Technology, Nanjing, Jiangsu 210094, China;
    2. School of Computer Science, Nanjing University of Science and Technology, Nanjing, Jiangsu 210094, China
  • Received:2011-11-28 Revised:2012-05-09 Online:2012-09-25 Published:2012-09-25
    • Supported by:
    • National Natural Science Foundation of China (No.61071146, No.61101194, No.61101198); Natural Science Foundation of Jiangsu Province,  China (No.BK2011701); Research Fund for the Doctoral Program of Higher Education of China (No.20113219120024)

摘要: 结合预估和残差补偿的递归重建算法是一种有效的压缩感知视频图像重建算法.针对现有算法中'预估’精度不高的问题,本文基于视频序列中相邻图像的内容相似性和单幅图像的非局部自相似性,分析了相邻图像局部图像块的相似匹配性,并以此作为视频图像的相关性先验,提出了一种基于局部自回归模型的图像预估重建算法.预估算法中当前图像像素点的自回归参数由参考图像中相似图像块的灰度信息通过学习获得.实验结果表明,与同类算法相比,本文预估算法所对应的递归重建算法可获得更高质量的视频图像重建结果.

关键词: 压缩感知, 视频图像, 递归重建, 自回归模型, 残差补偿

Abstract: The approaches combining the prediction and residual compensation can be used to reconstruct the compressed sensing video images recursively.In order to improve the precision of the existing prediction schemes,this paper proposes an image prediction algorithm based on the local autoregressive model.Before that,this paper firstly analyses the similarity of local image patches in consecutive images based on the content similarity of two consecutive images and the non-local similarity of each single image,and then takes this similarity as the correlation prior information to estimate the autoregressive parameters of current image.Compared to the existing related algorithms,the recursive reconstruction algorithm exploiting the proposed prediction scheme can achieve higher video images reconstruction performance.

Key words: compressed sensing, video images, recursive reconstruction, local autoregressive model, residual compensation

中图分类号: