发表机构
University of Science and Technology of China; Alibaba Group（中国科学技术大学; 阿里巴巴集团）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
ResARC通过扩散变换器生成量化残差并学习压缩生成残差，在解码端补偿两种残差，实现超低比特率下高保真图像压缩。
AI 中文摘要
渐进式自回归图像编解码器通过将连续潜变量量化为离散标记，传输从粗到细的前缀标记，并在解码器端生成剩余的后缀标记，为生成式压缩提供了一种有吸引力的范式。然而，其重建质量从根本上受到该流程中引入的两种残差的限制：量化残差（源于离散标记化过程中的信息损失）和生成残差（源于后缀标记的不完美自回归生成）。为解决这些限制，我们提出了ResARC，一种残差感知的自回归编解码器，在解码器端显式补偿这两种残差。具体而言，我们使用以自回归解码上下文为条件的扩散变换器生成量化残差，且无需额外的边信息。同时，我们在编码器端计算生成残差，并采用学习到的生成残差编解码器（Generation Residual Codec）对其进行高效压缩和传输，用于解码器端校正。恢复的残差随后与重建的潜表示集成，并通过适配的VAE解码器进行解码。大量实验表明，ResARC在超低比特率范围内实现了与领先生成式编解码器相当的感知相似性，同时显著提升了分布保真度。代码和模型将很快发布。
英文摘要
Progressive autoregressive image codecs provide an appealing paradigm for generative compression by quantizing continuous latents into discrete tokens, transmitting coarse-to-fine prefix tokens and generating the remaining suffix tokens at the decoder. However, their reconstruction quality is fundamentally limited by two residuals introduced along this pipeline: the quantization residual, arising from information loss during discrete tokenization, and the generation residual, resulting from imperfect autoregressive generation of the suffix tokens. To address these limitations, we introduce ResARC, a residual-aware autoregressive codec that explicitly compensates for both residuals at the decoder. Specifically, we generate the quantization residual with a diffusion transformer conditioned on the autoregressive decoding context, while requiring no additional side information. In parallel, we compute the generation residual at the encoder and employ a learned Generation Residual Codec to efficiently compress and transmit it for decoder-side correction. The recovered residuals are then integrated with the reconstructed latent representation and decoded through an adapted VAE decoder. Extensive experiments demonstrate that ResARC achieves competitive perceptual similarity while substantially improving distributional fidelity over leading generative codecs across the ultra-low bitrate regime. Code and models will be released soon.
