A Hybrid $L_2-L_p$ Variational Model for Single Low-light Image Enhancement with Bright Channel Prior

Fu Gang, Duan Lian, Xiao Chunxia

Note: if you have any question, please contact me by Email: xyzgfu@gmail.com

Introduction

In this paper, we consider and study the norm variable and propose a hybrid $L_{2}-L_{p}$ variational model with bright channel prior based on Retinex to decompose an observed image into a reflectance layer and an illumination layer. Different from the existing methods, our proposed model can preserve the reflectance layer with more fine details while enforcing the illumination layer to be texture-less, avoiding the texture-copy problem. Moreover, for solving our non-linear optimization, we adopt an alternating minimization scheme to find the optimal. Finally, we test our algorithm on a large number of images and the experimental results illustrate that the proposed method has achieved the better result than other state-of-the-art methods both qualitatively and quantitatively.

If you use the code for your research, please cite our paper as follows:

@inproceedings{fu-2019-hybrid-l,
author =       {Fu Gang, Duan Lian, Xiao Chunxia},
title =        {A Hybrid $L_2-L_p$ Variational Model for Single Low-light Image Enhancement with Bright Channel Prior},
booktitle =    {IEEE International Conference on Image Processing (ICIP)},
year =         {2019},
}

Execution platform

I have tested this code on Linux (Manjaro) system with Matlab 2014b (2017a), but it shoud work on other version of Matlab. For any question about the code, please contact me by xyzgfu@gmail.com.