Salvato in:
| Autori principali: | , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | https://arxiv.org/abs/2604.10546 |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866910187024023552 |
|---|---|
| author | Jiang, Shiyin Long, Wei Han, Minghao Chen, Zhenghao Zhu, Ce Gu, Shuhang |
| author_facet | Jiang, Shiyin Long, Wei Han, Minghao Chen, Zhenghao Zhu, Ce Gu, Shuhang |
| contents | The rapid growth of visual data under stringent storage and bandwidth constraints makes extremely low-bitrate image compression increasingly important. While Vector Quantization (VQ) offers strong structural fidelity, existing methods lack a principled mechanism for joint rate-distortion (RD) optimization due to the disconnect between representation learning and entropy modeling. We propose RDVQ, a unified framework that enables end-to-end RD optimization for VQ-based compression via a differentiable relaxation of the codebook distribution, allowing the entropy loss to directly shape the latent prior. We further develop an autoregressive entropy model that supports accurate entropy modeling and test-time rate control. Extensive experiments demonstrate that RDVQ achieves strong performance at extremely low bitrates with a lightweight architecture, attaining competitive or superior perceptual quality with significantly fewer parameters. Compared with RDEIC, RDVQ reduces bitrate by up to 75.71% on DISTS and 37.63% on LPIPS on DIV2K-val. Beyond empirical gains, RDVQ introduces an entropy-constrained formulation of VQ, highlighting the potential for a more unified view of image tokenization and compression. The code will be available at https://github.com/CVL-UESTC/RDVQ. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_10546 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Differentiable Vector Quantization for Rate-Distortion Optimization of Generative Image Compression Jiang, Shiyin Long, Wei Han, Minghao Chen, Zhenghao Zhu, Ce Gu, Shuhang Computer Vision and Pattern Recognition The rapid growth of visual data under stringent storage and bandwidth constraints makes extremely low-bitrate image compression increasingly important. While Vector Quantization (VQ) offers strong structural fidelity, existing methods lack a principled mechanism for joint rate-distortion (RD) optimization due to the disconnect between representation learning and entropy modeling. We propose RDVQ, a unified framework that enables end-to-end RD optimization for VQ-based compression via a differentiable relaxation of the codebook distribution, allowing the entropy loss to directly shape the latent prior. We further develop an autoregressive entropy model that supports accurate entropy modeling and test-time rate control. Extensive experiments demonstrate that RDVQ achieves strong performance at extremely low bitrates with a lightweight architecture, attaining competitive or superior perceptual quality with significantly fewer parameters. Compared with RDEIC, RDVQ reduces bitrate by up to 75.71% on DISTS and 37.63% on LPIPS on DIV2K-val. Beyond empirical gains, RDVQ introduces an entropy-constrained formulation of VQ, highlighting the potential for a more unified view of image tokenization and compression. The code will be available at https://github.com/CVL-UESTC/RDVQ. |
| title | Differentiable Vector Quantization for Rate-Distortion Optimization of Generative Image Compression |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2604.10546 |