Low-light Image Enhancement Based on Weighted Adaptive Guided Filter

Abstract

It is known that filtering process always yields blurring and loss of details in image processing, especially in the case of noisy. In this paper, a new weighted adaptive guided filter (WAGIF) is proposed to estimate illumination component. Specifically, the proposed WAGIF consists of three steps: 1) Local skewness is employed to distinguish the flat areas and texture areas of the underlying image. 2) An adaptive regularization parameter is designed to filter the image adaptively according to the skewness. 3) The filtered results are weighted and aggregated to output the final image. After obtaining the illumination component, the Retinex model is used to obtain the reflection component. Furthermore, the illumination component is processed by sigmoid function and the reflection component is enhanced by gamma correction. Finally, the illumination component and the reflection component are reconstructed to obtain an image with high contrast and rich details. The experimental results show that the proposed method is better than the existing methods in terms of denoising performance and structure preservation for low-light image enhancement.