Super-High-Fidelity Image Compression via Hierarchical-ROI and Adaptive Quantization
This work addresses visual quality issues in image compression for applications requiring high-fidelity reconstruction of faces and text, representing an incremental improvement over existing methods.
The paper tackled the problem of blurring and deformation in learned image compression at low bit rates, especially for small faces and text, by combining MSE-based and generative models with a hierarchical ROI approach and adaptive quantization, achieving better visual quality with lower bit rates, such as 0.7X bits of HiFiC and 0.5X bits of BPG.
Learned Image Compression (LIC) has achieved dramatic progress regarding objective and subjective metrics. MSE-based models aim to improve objective metrics while generative models are leveraged to improve visual quality measured by subjective metrics. However, they all suffer from blurring or deformation at low bit rates, especially at below $0.2bpp$. Besides, deformation on human faces and text is unacceptable for visual quality assessment, and the problem becomes more prominent on small faces and text. To solve this problem, we combine the advantage of MSE-based models and generative models by utilizing region of interest (ROI). We propose Hierarchical-ROI (H-ROI), to split images into several foreground regions and one background region to improve the reconstruction of regions containing faces, text, and complex textures. Further, we propose adaptive quantization by non-linear mapping within the channel dimension to constrain the bit rate while maintaining the visual quality. Exhaustive experiments demonstrate that our methods achieve better visual quality on small faces and text with lower bit rates, e.g., $0.7X$ bits of HiFiC and $0.5X$ bits of BPG.