Search papers, labs, and topics across Lattice.
This paper introduces AIGS-Net, a lightweight architecture designed for fast low-light image enhancement by utilizing an adaptive 2D Gaussian Splatting illumination field. The model dynamically adjusts the opacity of Gaussian basis functions based on the input image's luminance statistics, enabling effective spatially varying illumination compensation while maintaining a minimal parameter count. Experimental results demonstrate that AIGS-Net significantly enhances detail recovery and color fidelity on benchmark datasets, achieving a remarkable balance between enhancement quality and computational efficiency with only about 40 learnable parameters.
AIGS-Net achieves superior low-light image enhancement with only 40 learnable parameters, redefining the trade-off between quality and efficiency in image processing.
Existing low-light image enhancement methods often face a bottleneck between the representation capacity of illumination-field modeling and computational complexity. To address this issue, this paper proposes an Adaptive Illumination Gaussian Splatting Network (AIGS-Net), an ultra-lightweight architecture for fast low-light enhancement. Unlike conventional static priors, AIGS-Net constructs an input-adaptive 2D Gaussian Splatting illumination field. The opacity of Gaussian basis functions is dynamically modulated by relative luminance statistics of the input image, and spatially varying illumination compensation is rendered through ordered alpha compositing. To guide adaptive illumination compensation efficiently, a zero-parameter nonlinear multiscale contextual encoding module is introduced to extract low-frequency structures and local contrast cues without additional convolutional weights. To suppress noise amplification and sensor-induced color bias, AIGS-Net integrates noise-mask estimation, locked single-channel Gamma mapping, cross-channel consistency regularization, and target color-alignment constraints. Experiments on LOL and LSRW benchmarks show that AIGS-Net improves detail recovery and color fidelity while requiring only approximately 40 learnable parameters, achieving an effective trade-off between enhancement quality and extreme inference efficiency.
CMU ML