Flow Straight to Reality: Perceptually Consistent Flow Matching for Efficient Image Restoration

2026-08-11

AI Summary

This work addresses the inherent trade-off in image restoration between distortion minimization and perceptual quality: reducing pixel-wise error often leads to over-smoothing, while enhancing perceptual realism may introduce structural inaccuracies. To reconcile this conflict, the authors propose PCFlow, a framework that directly maps degraded images to sharp targets via continuous flow matching in a latent space, jointly optimizing both distortion and perceptual objectives. Key innovations include a latent consistency-aware perceptual loss (LCPL) to regularize the velocity field, a conflict-free gradient projection mechanism to stabilize multi-objective optimization, and a lightweight, purely convolutional backbone enabling efficient few-step inference. The method achieves competitive performance across diverse restoration tasks while significantly reducing computational overhead.

Abstract

Image restoration is fundamentally constrained by the tradeoff between distortion and perception: minimizing pixel-wise error yields over-smoothed results, whereas optimizing for perceptual realism often introduces structural deviations. Recent approaches attempt to balance this tradeoff via posterior sampling or multi-stage generative pipelines, yet remain computationally expensive and architecturally complex. To overcome these limitations, we propose PCFlow (Perceptually Consistent Flow Matching), a unified framework that directly parameterizes a continuous transport from degraded observations to clean targets, jointly optimizing distortion and perceptual quality. While its latent consistency flow objective drives stable and efficient few-step inference, a Latent Consistency Perceptual Loss (LCPL) imposes semantic constraints directly on the guiding velocity field, steering the dynamics toward visually sharp data manifolds. Furthermore, recognizing the inherent conflict between structural and perceptual consistencies, we integrate a conflict-free gradient projection strategy to stabilize the multi-objective optimization landscape. Combined with lightweight, convolution-only backbone, PCFlow achieves competitive performance across diverse restoration tasks at a fraction of traditional computational costs.

Problem

• image restoration
• distortion-perception tradeoff
• perceptual consistency
• structural deviation
• computational efficiency

Innovation

• flow matching
• perceptual consistency
• image restoration
• latent consistency
• multi-objective optimization