一歩で画像復元する新フレームワーク、精度と速度を両立
NullFlow: One-Step Generative Reconstruction

- 生成フローを測定値一致部分空間に制約し、別途の補正不要
- 1回のネットワーク評価でサンプリング可能、従来の数百回から劇減
- 画像修復タスクで最先端の拡散モデルと同等性能
我们提出NullFlow,一种原理严谨的一步生成式图像重建框架。核心思想是将生成流限制在与测量值一致的子空间内。由于流始终在此子空间中,无需额外的数据保真度校正,区别于现有求解器。NullFlow通过学习流的平均速度,在单次网络评估中完成采样,避免了传统流匹配方法的逐步积分。我们证明,该约束流的平均速度可导出一个训练目标,其全局最小值即为一步后验采样器。在图像修复任务中,NullFlow达到与最先进扩散求解器相当的性能,同时将推理过程从数百次网络评估减少至一次。
原文摘要 · Abstract (English)
We propose NullFlow, a principled framework for one-step generative image reconstruction. Our key idea is to confine the generative flow to a measurement-consistent subspace. Because the flow never leaves this subspace, NullFlow needs no separate data-fidelity corrections, unlike existing solvers. NullFlow samples in a single network evaluation by learning the flow's average velocity, avoiding the step-by-step integration of traditional flow matching methods. We prove that the average velocity of this constrained flow yields a training objective whose global minimizer is a one-step posterior sampler. We show on image inpainting that NullFlow matches state-of-the-art diffusion solvers while cutting inference from hundreds of network evaluations to one.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品;中文卡片由大模型生成,请以原文为准。