作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程

• 图形图像处理 • 上一篇    下一篇

基于新边缘指导插值的迭代反投影超分辨率重建算法

陶志强,李海林,张红兵   

  1. (南京航空航天大学 电子信息工程学院,南京 210016)
  • 收稿日期:2015-05-28 出版日期:2016-06-15 发布日期:2016-06-15
  • 作者简介:陶志强(1990-),男,硕士研究生,主研方向为图形图像处理、数字信号处理;李海林,讲师;张红兵,硕士研究生。
  • 基金资助:
    国家自然科学基金资助项目(61371170);中央高校基本科研业务费专项基金资助项目(NJ20140010);南京航空航天大学研究生创新基地(实验室)开放基金资助项目(kfjj201428)。

Iterative Back Projection Super Resolution Reconstruction Algorithm Based on New Edge Directed Interpolation

TAO Zhiqiang,LI Hailin,ZHANG Hongbing   

  1. (College of Electronic and Information Engineering,Nanjing University of Aeronautics and Astronautics,Nanjing 210016,China)
  • Received:2015-05-28 Online:2016-06-15 Published:2016-06-15

摘要: 基于改进Keren配准算法的迭代反投影(IBP)超分辨率重建算法,使用双线性插值方法获得高分辨率图像的初始估计,导致重建图像产生边缘锯齿效应。针对该问题,提出一种基于新边缘指导插值(NEDI)的IBP超分辨率重建算法。利用低分辨率图像与高分辨率图像的局部协方差间的几何对偶性,通过计算低分辨率图像各像素点的局部协方差系数,得到高分辨率图像待插值像素点的值。实验结果表明,该算法能够有效减小边缘锯齿,提高峰值信噪比,降低均方根误差,并改善图像的主观视觉效果。

关键词: 超分辨率重建, 迭代反投影, 新边缘指导插值, 峰值信噪比, 均方根误差

Abstract: The Iterative Back Projection(IBP) Super Resolution Reconstruction(SRR) algorithm based on improved Keren registration method uses bilinear interpolation method to get the initial estimations of high-resolution images,which leads to the sawtooth in the edge of the reconstructed image.To solve this problem,an IBP super resolution reconstruction based on New Edge Directed Interpolation(NEDI) is proposed.The NEDI method computes local covariance coefficients of a low-resolution image and uses these covariance estimations to adapt the interpolation at a higher resolution based on the geometric duality between the local covariance of the low-resolution image and that of the high-resolution image.Experimental result indicates that the proposed method can reduce the edge sawtooth,increase Peak Signal to Noise Ratio(PSNR),reduce Root Mean Squared Error(RMSE) and improve the subjective visual effect of the image.

Key words: Super Resolution Reconstruction(SRR), Iterative Back Projection(IBP), New Edge Directed Interpolation(NEDI), Peak Signal to Noise Ratio(PSNR), Root Mean Squared Error(RMSE)

中图分类号: