AI 中文总结
MeanSR是一种单步感知超分辨率方法，通过学习LR条件平均速度场并引入阶段感知时间采样策略，在CLIPIQA等指标上优于CTMSR，同时降低计算成本与延迟，提升了重建质量。
AI 中文摘要
基于扩散的超分辨率（SR）方法可实现出色的感知质量，但需要高昂的迭代去噪成本。现有的单步蒸馏方法虽能减少推理时间，但依赖昂贵的预训练教师模型；而CTMSR通过PF-ODE一致性训练避免了蒸馏，却未明确建模从低分辨率（LR）输入到高分辨率（HR）图像的恢复动态。我们提出MeanSR，这是一种单步感知SR方法，学习LR条件下的平均速度场，以直接捕捉从退化或带噪输入到合理HR输出的有限时间转换。我们进一步重新构建用于平均速度生成的分布轨迹匹配，并引入阶段感知时间采样策略以改进轨迹学习。在合成与真实基准上的实验表明，MeanSR在CLIPIQA、MUSIQ和MANIQA指标上优于CTMSR，同时大幅降低了FLOPs和推理延迟，还能重建更清晰的结构、更真实的纹理，且感知伪影更少。
英文摘要
Diffusion-based super-resolution (SR) achieves strong perceptual quality but requires costly iterative denoising. Existing one-step distillation methods reduce inference time but depend on expensive pretrained teachers, whereas CTMSR avoids distillation through PF-ODE consistency training yet does not explicitly model the restoration dynamics from low-resolution (LR) inputs to high-resolution (HR) images. We propose MeanSR, a one-step perceptual SR method that learns an LR-conditioned average velocity field to directly capture the finite-time transition from degraded or noisy inputs to plausible HR outputs. We further reformulate distribution trajectory matching for average-velocity generation and introduce a Stage-Aware Temporal Sampling strategy to improve trajectory learning. Experiments on synthetic and real-world benchmarks show that MeanSR outperforms CTMSR on CLIPIQA, MUSIQ, and MANIQA while substantially reducing FLOPs and inference latency. MeanSR also reconstructs sharper structures and more realistic textures with fewer perceptual artifacts.
Comments7 pages, 8 figures. Submitted to AAAI 2027
