FS-NCSR: Increasing Diversity of the Super-Resolution Space via Frequency Separation and Noise-Conditioned Normalizing Flow.
Ki-Ung SongDongseok ShimKang-Wook KimJae-Young LeeYounggeun KimPublished in: CVPR Workshops (2022)
Keyphrases
- super resolution
- low resolution
- high resolution
- iterative reconstruction algorithm
- image reconstruction
- image restoration
- motion estimation
- low resolution images
- imaging process
- high quality
- motion blur
- degraded images
- sharp edges
- image super resolution
- multi frame
- high resolution images
- higher resolution
- resolution enhancement
- super resolution reconstruction
- single image super resolution
- image resolution
- super resolved
- gaussian noise
- super resolution algorithm
- depth map
- combining information from multiple
- video super resolution
- feature selection
- single frame
- computational complexity
- feature space
- light field
- spatial resolution
- image patches
- frequency domain