Compressed Image Super Resolution

Compressed image super-resolution (CSR) aims to enhance the resolution of images already compressed, addressing the combined challenges of low resolution and compression artifacts. Recent research focuses on developing sophisticated neural network architectures, such as transformer-based models (e.g., Swin Transformers) and those incorporating novel scanning strategies, to effectively model complex contextual information and mitigate these distortions across various compression codecs. This field is significant because it improves the quality of images and videos transmitted and stored in compressed formats, impacting applications ranging from streaming services to virtual reality.