← 피드로
[Submitted on 5 Dec 2025 (v1), last revised 25 Sep 2026 (this version, v4)]
Abstract:Retinex-based low-light image enhancement benefits from separating reflectance and illumination, yet recent generative approaches often rely on iterative sampling and are difficult to deploy under strict latency budgets. Consistency models offer a natural route to one-step restoration, but direct adaptation to Retinex-factorized enhancement is unstable: one-step inference is evaluated at the high-noise endpoint, whereas standard training schedules provide little supervision there, and temporal self-consistency alone does not determine the correct conditional target. We propose Consist-Retinex, which first uses a Retinex Transformer Decomposition Network (TDN) to obtain paired reflectance and illumination maps, then trains two conditional consistency models with a Retinex-aware dual objective and adaptive noise-emphasized fixed-point sampling. The dual objective combines trajectory consistency with paired ground-truth component alignment, while the sampling rule concentrates supervision near the inference endpoint without discarding full-range noise coverage. We further provide an endpoint error bound, an anchoring-propagation result, and a high-noise sample-allocation analysis that explain why endpoint supervision and temporal consistency are complementary for one-step Retinex enhancement. Experiments on paired and unpaired low-light benchmarks show that Consist-Retinex obtains the best VE-LOL-L scores among the compared methods under one-step inference and remains competitive on LOL, with substantially reduced sampling and consistency-stage training cost in the reported setup.
Submission history
From: Xu Jian [view email]
[v1]
Fri, 5 Dec 2025 13:44:19 UTC (16,904 KB)
[v2]
Wed, 29 Apr 2026 12:19:25 UTC (16,917 KB)
[v3]
Mon, 4 May 2026 12:25:04 UTC (17,635 KB)
[v4]
Fri, 25 Sep 2026 13:01:31 UTC (11,685 KB)
추출 본문 · 출처: arxiv.org · https://arxiv.org/abs/2512.08982