Beyond Fixed Inference: Quantitative Flow Matching for Adaptive Image Denoising

Abstract

Diffusion and flow-based generative models have shown strong potential for image restoration. However, image denoising under unknown and varying noise conditions remains challenging, because the learned vector fields may become inconsistent across different noise levels, leading to degraded restoration quality under mismatch between training and inference. To address this issue, we propose a quantitative flow matching framework for adaptive image denoising. The method first estimates the input noise level from local pixel statistics, and then uses this quantitative estimate to adapt the inference trajectory, including the starting point, the number of integration steps, and the step-size schedule. In this way, the denoising process is better aligned with the actual corruption level of each input, reducing unnecessary computation for lightly corrupted images while providing sufficient refinement for heavily degraded ones. By coupling quantitative noise estimation with noise-adaptive flow inference, the proposed method improves both restoration accuracy and inference efficiency. Extensive experiments on natural, medical, and microscopy images demonstrate its robustness and strong generalization across diverse noise levels and imaging conditions.

Figures (11)

• Figure 1: Qualitative denoising results of the proposed method under different Gaussian noise levels. For each example, the left image is the noisy input and the right image is the corresponding denoised result. The noise standard deviation $\sigma$ is shown in the upper-left corner of each noisy image. (a)--(c) correspond to medium-, high-, and very high-noise conditions, respectively.
• Figure 2: Conceptual comparison of denoising mechanisms under different noise conditions. (a) Traditional network-based model. (b) Flow-based generative model with a fixed inference trajectory. (c) Conditional flow matching model guided by the noisy observation. (d) Ideal noise-aware model with trajectories better matched to different noise levels. Here, $x_0$ denotes the clean image, $x_T$ denotes the noisy initial state, and $x_1$, $x_2$, and $x_3$ represent inputs under high-, medium-, and low-noise conditions, respectively.
• Figure 3: Overview of the proposed method. (a) Overall pipeline, including noise-level estimation, computation of the adaptive starting time and inference grid, and reverse flow integration for denoising. (b) Intermediate denoising results at different time points during adaptive inference.
• Figure 4: Pipeline of the proposed quantitative noise estimation method. (a) Illustration of block-wise pixel differences in noise-free and noisy images after partitioning into non-overlapping $2\times2$ patches. (b) Estimation of the global noise level using range and middle-range statistics, followed by calibration, random fusion, and sample averaging.
• Figure 5: Qualitative denoising comparisons on the BSDS 500 dataset. From left to right: noisy input, supervised baseline DVT yang2024denoising, self-supervised baseline MASH chihaoui2024mash, flow-based baseline DeltaFM stoica2025contrastive, conditional flow-matching baseline CE-CFM cheng2025c2ot, the proposed method, and the ground-truth reference. From top to bottom, the Gaussian noise standard deviations are 0.22, 0.51, 0.61, and 1.00, respectively. The PSNR/SSIM values for each result are shown below the corresponding image.