发表机构
Shanghai Jiao Tong University（上海交通大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
PixelIR是基于像素空间图像残差流匹配的保真度-感知解耦Real-ISR框架，经蒸馏后单步重建，在RealSR等数据集指标领先，参数、计算量与延迟低，实现三者平衡。
AI 中文摘要
真实世界图像超分辨率（Real-ISR）旨在在退化观测的基础上保留结构，同时重建具有感知真实感的细节。然而，现有Real-ISR方法大多在共享网络内优化保真度和感知质量，导致两个目标在训练过程中相互干扰，难以控制两者的平衡。近期的单步方法减少了采样步数，但通常继承了这种耦合优化行为以及其多步前体方法昂贵的高分辨率骨干网络。我们认为，高效的Real-ISR不仅需要更短的采样轨迹，还需要专门对忠实重建和感知细节合成进行建模。基于这一见解，我们提出PixelIR，这是一个基于像素空间图像残差流匹配构建的保真度-感知解耦框架。PixelIR首先学习将退化观测映射到忠实重建的图像流，然后残差流从噪声中合成缺失的感知细节，无需重复重新学习或覆盖完整的重建方案。我们进一步在由粗到细的金字塔架构中将教师模型蒸馏为面向部署的单步学生模型。大量实验表明，PixelIR在RealSR和DRealSR数据集上均达到领先的PSNR、SSIM和LPIPS指标。最终模型仅用32.9M参数、89.7G MACs和8.5ms延迟即可完成像素空间的单次评估重建，展现出强大的保真度-感知-效率平衡能力。
英文摘要
Real-world image super-resolution (Real-ISR) aims to preserve structures supported by the degraded observation while reconstructing perceptually realistic details. However, existing Real-ISR methods largely optimize fidelity and perceptual quality within a shared network, causing the two objectives to interfere throughout training and making their balance difficult to control. Recent one-step methods reduce sampling steps, yet often inherit both this coupled optimization behavior and the expensive high-resolution backbone of their multi-step predecessors. We argue that efficient Real-ISR requires not only a shorter sampling trajectory, but also specialized modeling of faithful reconstruction and perceptual detail synthesis. Based on this insight, we propose PixelIR, a fidelity-perception decoupling framework built upon pixel-space image-residual flow matching. PixelIR first learns an image flow that maps the degraded observation to a faithful reconstruction. Then, a residual flow synthesizes the missing perceptual details from noise without repeatedly relearning or overwriting the complete restoration solution. We further distill the teacher into a deployment-oriented one-step student within a coarse-to-fine pyramid architecture. Extensive experiments show that PixelIR achieves leading PSNR, SSIM, and LPIPS on both RealSR and DRealSR. The final model completes pixel-space restoration in a single evaluation with only 32.9M parameters, 89.7G MACs, and 8.5ms latency, demonstrating a strong practical fidelity-perception-efficiency balance.
