A Novel Low-Illumination Image Enhancement Method Based on Dual-Channel Prior

Low-illumination image enhancement is the important part of computer vision research, and defogging model is widely applied to image enhancement. However, there are still some problems in this model, such as the different depth of field of the window segmentation and blurring of edge details. In this paper, we propose a novel low-illumination image enhancement method based on dual-channel prior. First, the low-illumination image is divided into regions with the same depth of field in the local window using the superpixel segmentation method. Then the bright and dark channels of the image are obtained through the maximum/minimum filtering, and after the bright channel is introduced, the adaptive threshold method is used to obtain the atmospheric light value. The initial transmission rate is obtained by combining the dual-channel image with pixel-level image fusion and determining the optimal combination coefficient for mutual correction. The light map with clearer texture and brightness information is used as the guide map for multi-scale gradient domain guided filtering to refine the transmittance. Finally fast BM3D removes noise and obtains the final enhanced image. Experimental results prove that the method proposed in this paper has improved in removing halo artifacts and restoring detailed information.