A New Low Light Image Enhancement Based on the Image Degradation Model

In this paper, a new low light image enhancement is presented based on the image degradation model, which belongs to the Retinex-based strategy. Our method intends to enhance a low light image by estimating its reflectance and illumination simultaneously. In addition, our method considers the noise in the dark regions of low light images. We explore the prior knowledge of the noise and related information of the reflectance, and an alternating direction minimization algorithm is provided to optimize the objective function. To speed up the algorithm, we use the Fast Fourier Transformation. Compared with other state-of-the-arts, the proposed method can provide better enhanced result from both subjective and objective aspects.