Deep Edge-Aware Post-Processing for JPEG Enhancement: CNN-Based Artifact Reduction and Image Quality Restoration
A CNN-based edge-aware artifact reduction framework (CNN-AR) is proposed that integrates an enhanced deep super-resolution (EDSR) backbone with a holistically nested edge detection (HED) guided loss, enabling superior artifact suppression while preserving fine structural details.