CISRDCNN: Super-resolution of compressed images using deep convolutional neural networks.
Honggang ChenXiaohai HeChao RenLinbo QingQizhi TengPublished in: CoRR (2017)
Keyphrases
- super resolution
- compressed images
- convolutional neural networks
- image reconstruction
- reconstructed image
- low resolution
- high resolution
- image compression
- visual quality
- image restoration
- low resolution images
- compression ratio
- image super resolution
- high quality
- bitstream
- spatial resolution
- high frequency
- motion blur
- compressed domain
- motion estimation
- discrete cosine transform
- dct domain
- peak signal to noise ratio
- high resolution images
- depth map
- subband
- image segmentation