Low-Complexity Single-Image Super-Resolution based on Nonnegative Neighbor Embedding


The paper
M. Bevilacqua, A. Roumy, C. Guillemot and ML. Alberi.
``Low-Complexity Single-Image Super-Resolution based on Nonnegative Neighbor Embedding'',
BMVC 2012. (PDF) (Abstract-PDF)


Comparison of our SR method to other SR methods.
We compare our algorithm with
[Chang]Hong Chang, Dit-Yan Yeung, and Yimin Xiong. Super-Resolution Through Neighbor Embedding. In 2004 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR), volume 1, pages 275–282, 2004.
[Glasner] Daniel Glasner, Shai Bagon, and Michal Irani. Super-Resolution from a Single Image. In 2009 IEEE 12th International Conference on Computer Vision (ICCV), pages 349– 356, 10 2009.
[Tang]Yi Tang, Pingkun Yan, Yuan Yuan, and Xuelong Li. Single-image super-resolution via local learning. International Journal of Machine Learning and Cybernetics, 2:15–23, 2011.
Our algorithm Chang et al. Glasner et al. Tang et al. Lin. Regr. + F2
Image Scale PSNR Time PSNR Time PSNR Time PSNR Time PSNR Time
baby 2 34.64 58 33.42 339 34.66 4083 33.72 425 34.76 30
bird 2 34.69 18 32.94 110 34.42 406 33.31 132 34.91 9
butterfly 2 27.54 17 25.90 77 26.83 265 26.05 82 27.66 7
head 2 32.88 18 32.34 145 32.68 367 32.43 151 32.88 8
woman 2 30.91 15 29.43 114 30.61 410 29.64 128 31.01 8
baby 3 32.44 27 31.00 116 32.94 2188 31.47 111 32.59 12
bird 3 31.37 9 29.71 47 32.16 281 30.07 42 31.57 5
butterfly 3 24.31 9 22.58 34 25.66 232 22.72 25 24.47 4
head 3 31.46 12 30.82 68 31.69 370 30.95 54 31.55 4
woman 3 27.98 12 26.45 37 28.79 248 26.66 37 28.12 4
baby 4 30.62 22 29.27 86 31.41 4381 29.70 81 30.76 13
bird 4 28.99 6 27.37 21 30.07 475 27.84 22 29.12 3
butterfly 4 22.05 7 20.50 18 23.94 315 20.61 13 22.06 3
head 4 30.26 6 29.57 26 30.86 379 29.83 28 30.43 3
woman 4 25.66 5 24.25 17 26.79 401 24.46 20 25.69 3

Results (PSNR and running time in sec.) for different images.