Image Resolution Enhancer using Deep Learning
Harsh Mittal, Vaibhav Rai, Swaraj Sonawane, Sneha Mhatre · 2022 International Conference on Applied Artificial Intelligence and Computing (ICAAIC) · 2022
Image Super-Resolution is a technique that is used to obtain high-resolution, realistic images from low-resolution input images. Deep learning algorithms such as SRCNN, ESRGAN, RDN, etc. have shown significant results in this field. But these algorithms at times vary in results. To solve this problem, this research study has proposed an image super-resolution by using Patch Extraction on Deep Learning Algorithm, in which the LR image is first divided into patches and then the algorithms like RDN and ESRGAN are applied. Comparing each patch from each algorithm based on PSNR values, the patch with the highest PSNR value will be selected. After picking up all the patches for that image, it will be reconstructed and hence the super-resolution image will be obtained as the output.