Detail preserving noise aware retinex model for low light image enhancement

Low-light image enhancement is a challenging task for human visual perception and high-quality image display due to the limited visibility in dark environments. Images captured in under-exposed environments often suffer from degradation issues of poor image quality, low signal-to-noise ratio, unclear details in dark areas, and overall low brightness. This article proposes a Retinex variational decomposition-based detail-preserving noise suppression model to address quality degradation issues of the images captured in under-exposed environments. First, the Retinex variational decomposition model effectively estimates the illumination and reflectance to facilitate high-quality image enhancement. Then, Weighted transformation is employed to adjust the illumination coefficients to improve overall visual quality. Finally, a bilateral non-iterative filter is used to suppress noise and illumination map estimation errors while preserving structural edges. Comprehensive experiments show that the proposed model performs better than existing enhancement methods both visually and quantitatively.