Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

MambaLLIE: Implicit Retinex-Aware Low Light Enhancement with Global-then-Local State Space

Jiangwei Weng, Zhiqiang Yan, Ying Tai, Jianjun Qian, Jian Yang, Jun Li

TL;DR

MambaLLIE introduces a global-then-local state-space framework for low-light image enhancement that combines a Local-Enhanced State Space Module with an implicit Retinex-aware Selective Kernel. By integrating LayerNorm, augmented illumination priors, and a Retinex-guided feature selection mechanism, the method models global long-range dependencies while preserving local detail. Across five benchmark datasets and real-world tests, it consistently outperforms CNN/Transformer-based methods and prior Retinex approaches, with substantial PSNR gains and improved perceptual quality. The approach advances LLIE by unifying state-space modeling with Retinex priors, offering robust enhancement in challenging lighting and preserving color fidelity, with potential extensions to video and other low-level vision tasks.

Abstract

Recent advances in low light image enhancement have been dominated by Retinex-based learning framework, leveraging convolutional neural networks (CNNs) and Transformers. However, the vanilla Retinex theory primarily addresses global illumination degradation and neglects local issues such as noise and blur in dark conditions. Moreover, CNNs and Transformers struggle to capture global degradation due to their limited receptive fields. While state space models (SSMs) have shown promise in the long-sequence modeling, they face challenges in combining local invariants and global context in visual data. In this paper, we introduce MambaLLIE, an implicit Retinex-aware low light enhancer featuring a global-then-local state space design. We first propose a Local-Enhanced State Space Module (LESSM) that incorporates an augmented local bias within a 2D selective scan mechanism, enhancing the original SSMs by preserving local 2D dependency. Additionally, an Implicit Retinex-aware Selective Kernel module (IRSK) dynamically selects features using spatially-varying operations, adapting to varying inputs through an adaptive kernel selection process. Our Global-then-Local State Space Block (GLSSB) integrates LESSM and IRSK with LayerNorm as its core. This design enables MambaLLIE to achieve comprehensive global long-range modeling and flexible local feature aggregation. Extensive experiments demonstrate that MambaLLIE significantly outperforms state-of-the-art CNN and Transformer-based methods. Project Page: https://mamballie.github.io/anon/

MambaLLIE: Implicit Retinex-Aware Low Light Enhancement with Global-then-Local State Space

TL;DR

MambaLLIE introduces a global-then-local state-space framework for low-light image enhancement that combines a Local-Enhanced State Space Module with an implicit Retinex-aware Selective Kernel. By integrating LayerNorm, augmented illumination priors, and a Retinex-guided feature selection mechanism, the method models global long-range dependencies while preserving local detail. Across five benchmark datasets and real-world tests, it consistently outperforms CNN/Transformer-based methods and prior Retinex approaches, with substantial PSNR gains and improved perceptual quality. The approach advances LLIE by unifying state-space modeling with Retinex priors, offering robust enhancement in challenging lighting and preserving color fidelity, with potential extensions to video and other low-level vision tasks.

Abstract

Recent advances in low light image enhancement have been dominated by Retinex-based learning framework, leveraging convolutional neural networks (CNNs) and Transformers. However, the vanilla Retinex theory primarily addresses global illumination degradation and neglects local issues such as noise and blur in dark conditions. Moreover, CNNs and Transformers struggle to capture global degradation due to their limited receptive fields. While state space models (SSMs) have shown promise in the long-sequence modeling, they face challenges in combining local invariants and global context in visual data. In this paper, we introduce MambaLLIE, an implicit Retinex-aware low light enhancer featuring a global-then-local state space design. We first propose a Local-Enhanced State Space Module (LESSM) that incorporates an augmented local bias within a 2D selective scan mechanism, enhancing the original SSMs by preserving local 2D dependency. Additionally, an Implicit Retinex-aware Selective Kernel module (IRSK) dynamically selects features using spatially-varying operations, adapting to varying inputs through an adaptive kernel selection process. Our Global-then-Local State Space Block (GLSSB) integrates LESSM and IRSK with LayerNorm as its core. This design enables MambaLLIE to achieve comprehensive global long-range modeling and flexible local feature aggregation. Extensive experiments demonstrate that MambaLLIE significantly outperforms state-of-the-art CNN and Transformer-based methods. Project Page: https://mamballie.github.io/anon/
Paper Structure (15 sections, 13 equations, 9 figures, 5 tables)

This paper contains 15 sections, 13 equations, 9 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Related work
  3. Methodology
  4. Preliminaries
  5. Overall Pipeline
  6. Global-then-Local State Space Block
  7. Experiments
  8. Benchmark Datasets and Implementation Details
  9. Main Results on Benchmarks.
  10. Real World Experimental Evaluation
  11. Ablation Study
  12. Limitation and Discussion
  13. Conclusion
  14. Broader Impact.
  15. More Results.

Figures (9)

  • Figure 1: The Effective Receptive Field (ERF) visualization ERP for SNR-Net SNR-Net, RetinexFormer RetinexFormer, MambaIR MambaIR and our MambaLLIE. A broader distribution of bright areas signifies a larger ERF. The receptive field of SNR-Net is large but messy, due to the SNR-aware mechanism, RetinexFormer achieves a larger receptive field of the central point, and MambaIR has the the global receptive field, but presents the limited local perception. Only our proposed MambaLLIE achieves a global perception ability outwards from central point and preserves the large local receptive field.
  • Figure 2: The overall pipeline of the proposed MambaLLIE. Our Global-then-Local State Space Block (GLSSB) integrates Local-enhanced state space module (LESSM) and implicit Retinex-aware selective kernel module (IRSK) with layer normalization as its core.
  • Figure 3: Qualitative comparison with previous methods on LOL-V2-real and LOL-V2-syn datasets. Our MambaLLIE effectively enhances the illumination and preserves the color.
  • Figure 4: Qualitative comparison with previous methods on SMID, SDSD-indoor and SDSD-outdoor datasets. Our MambaLLIE restore the texture and color under challenging degradation, such as the wooden bench and reflective glass.
  • Figure 5: Vision comparison of our MambaLLIE with recent SOTA methods.(a) Qualitative comparison on object detection, (b) A toy example of user study.
  • ...and 4 more figures