A low-light image enhancement framework based on hybrid multiscale decomposition and adaptive brightness adjustment model

Abstract

Under inadequate lighting conditions, low-light images often suffer from low contrast and poor visibility. However, many existing methods struggle to find a balance between detail enhancement, brightness adjustment, and noise suppression. To address these challenges, this paper proposes a hybrid multiscale decomposition and adaptive brightness adjustment model for low-light image enhancement. By combining local and global contrast enhancement techniques, an adaptive brightness adjustment algorithm is introduced to improve both the brightness and texture details. Furthermore, a hybrid multiscale decomposition model based on guided filtering and side window guided filters is designed to handle the intricate nature of image detail information, which divides the original image into three distinct layers: a base layer representing the background, a large-scale detail layer capturing prominent edge structures, and a small-scale detail layer preserving subtle texture details. To preserve key image details and enhance salient targets, fusion methods based on “exposure” functions and normalized arctan functions are employed. These methods ensure that weak details are preserved while suppressing noise artifacts. Qualitative and quantitative experimental results conducted on public datasets demonstrate that the proposed method surpasses state-of-the-art approaches in terms of detail enhancement, brightness adjustment, and noise suppression.