上海交通大学学报(英文版) ›› 2015, Vol. 20 ›› Issue (1): 26-31.doi: 10.1007/s12204-015-1583-1

• • 上一篇    下一篇

Single Image Super-Resolution Method via Refined Local Learning

TANG Song-ze*(唐松泽), XIAO Liang (肖亮), LIU Peng-fei (刘鹏飞)   

  1. (School of Computer Science and Technology, Nanjing University of Science and Technology, Nanjing 210094, China)
  • 出版日期:2015-02-28 发布日期:2015-03-10
  • 通讯作者: TANG Song-ze*(唐松泽) E-mail:ts198708@163.com

Single Image Super-Resolution Method via Refined Local Learning

TANG Song-ze*(唐松泽), XIAO Liang (肖亮), LIU Peng-fei (刘鹏飞)   

  1. (School of Computer Science and Technology, Nanjing University of Science and Technology, Nanjing 210094, China)
  • Online:2015-02-28 Published:2015-03-10
  • Contact: TANG Song-ze*(唐松泽) E-mail:ts198708@163.com

摘要:

In this paper, we propose a refined local learning scheme to reconstruct a high resolution (HR) face image from a low resolution (LR) observation. The contribution of this work is twofold. Firstly, multi-direction gradient features are extracted to search the nearest neighbors for each image patch, then the non-negative matrix factorization (NMF) is used to reduce the complexity in weight calculation, and the initial HR embedding is estimated from the training pairs by preserving local geometry. Secondly, a global reconstruction constraint and post-processing by non-local filtering is incorporated into super-resolution (SR) reconstruction process to reduce the image artifacts and further improve the image visual quality. Experimental results show that the proposed algorithm improves the SR performance both in subjective and objective assessments compared with several existing methods.

关键词: refined local learning, neighbor embedding, multi-direction, non-negative matrix factorization (NMF), post-processing

Abstract:

In this paper, we propose a refined local learning scheme to reconstruct a high resolution (HR) face image from a low resolution (LR) observation. The contribution of this work is twofold. Firstly, multi-direction gradient features are extracted to search the nearest neighbors for each image patch, then the non-negative matrix factorization (NMF) is used to reduce the complexity in weight calculation, and the initial HR embedding is estimated from the training pairs by preserving local geometry. Secondly, a global reconstruction constraint and post-processing by non-local filtering is incorporated into super-resolution (SR) reconstruction process to reduce the image artifacts and further improve the image visual quality. Experimental results show that the proposed algorithm improves the SR performance both in subjective and objective assessments compared with several existing methods.

Key words: refined local learning, neighbor embedding, multi-direction, non-negative matrix factorization (NMF), post-processing

中图分类号: