Computer Engineering and Applications ›› 2013, Vol. 49 ›› Issue (23): 186-190.

Previous Articles     Next Articles

Single image super-resolution combined with learning algorithm

HUANG Quanliang1, LIU Shuiqing1, SUN Jinhai2, CHEN Ke1   

  1. 1.Department of Electronics and Information Engineering, Huazhong University of Science and Technology, Wuhan 430074, China
    2.Science and Technology on Electromagnetic Scattering Laboratory, Beijing 100854, China
  • Online:2013-12-01 Published:2016-06-12


黄全亮1,刘水清1,孙金海2,陈  柯1   

  1. 1.华中科技大学 电子与信息工程系,武汉 430074
    2.电磁散射重点实验室,北京 100854

Abstract: In order to reconstruct a high-resolution image from single highly blurred image, a super-resolution reconstruction method combined with adaptive regularization and learning algorithm is proposed. Based on local characteristics, the dynamic adaptive control progress of reconstruction method is achieved. Train set, prediction principle and searching progress of learning algorithm is optimized to depress the relativity of example image and improve the searching efficiency. Experimental results demonstrate the availability of the method by steps, and the improvement of reconstruction result.

Key words: image reconstruction, super-resolution, regularization, learning algorithm

摘要: 为从高度降质的单帧图像中重建出高分辨率图像,提出了一个结合自适应正则化与学习算法的超分辨率复原方法。该方法基于图像的局部特征,实现了正则化方法动态自适应控制过程,优化了学习算法中的训练集、预测原则和搜索过程,以降低基准图相关性要求、提高搜索效率。仿真实验分步论证了该方法的有效性,以及对复原效果的提升。

关键词: 图像复原, 超分辨率, 正则化, 学习算法