An Adaptive Low-illumination Image Enhancement Algorithm based on Weighted Least Squares Optimization

Abstract An adaptive low-illumination image enhancement algorithm based on the weighted least squares optimization is proposed to solve difficulty of detailed feature recognition in images that collected by visible light imaging equipment. First, converted from RGB channel LAB channel. Second, we use an edge-preserving smoothing operator coarsen smooth base layer and extract multi-scale details brightness Then, weight applied fusion detail features. Finally, Retinex performed obtain a ultimate enhanced image. Experiments result show this method has suitable visual clear details. In terms objective indicators, it good stable performance NIQE, TMQI, information entropy.