ILVMamba: Illumination-Aware Lightweight Visual Mamba Framework for Efficient High-Resolution Image Enhancement

Real-world image quality is often degraded by suboptimal lighting conditions, such as low light and vignetting. With advancements in sensor technology, there is an increasing demand for efficient algorithms capable of processing high-resolution images. However, existing approaches primarily focus on low-resolution data or require significant computational resources for high-resolution processing. To address these challenges, we propose a lightweight framework, ILVMamba, specifically designed for high-resolution image restoration under adverse lighting conditions. We introduce Illumination-aware Visual Mamba Blocks (IVMBs), which are developed based on the Retinex theory. These IVMBs efficiently utilize illumination information to enhance image feature representation. Additionally, IVMBs integrate Mamba2 to reduce model complexity while preserving reliable performance on large-scale images. We conduct extensive experiments on various high-resolution datasets, including UHD-LOL4K for low-light enhancement and VigSet for vignetting removal. The results indicate that ILVMamba demonstrates competitive performance on public high-resolution benchmarks while maintaining low model complexity. Additionally, our method also shows superior performance on public low-resolution datasets. <ext-link ext-link-type="uri" xlink:href="https://github.com/MingyuLiu1/ILVMamba.git" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/MingyuLiu1/ILVMamba.git</ext-link>