Seven ways to improve example-based single image super resolution
Radu Timofte, Rasmus Rothe, and Luc Van Gool
CVPR 2016 Submission ID 769
This supplementary material provides images, PSNR results, and running-times for the methods from the paper:
R. Timofte, R. Rothe, and L. Van Gool.
Seven ways to improve example-based single image super resolution.
In IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR 2016), June 2016, US.
[ bib |
Supplementary material |
L20 dataset ]
Seven ways to improve example-based single image super resolution are demonstrated in the paper.
Particularly, in this supplementary material,
A+B, A+A, A+C, and IA are shown to improve in quality over the compared methods and the A+ baseline.
We keep the settings from the paper and the evaluation procedure.
L20 dataset: a new dataset with 20 high resolution large images.
Note that it is a re-run, there might be some differences with respect to the results from the paper.
Please click on the PSNR result values from the links below to see the grayscale image output for most of the methods.
The RGB image outputs are obtained as in the paper by bicubically interpolating the chroma components of the LR input image.