Image Enhancement in Varying Light Conditions Based on Wavelet Transform

Muhammad Aamir,Ziaur Rehman,Yi-Fei Pu,Awais Ahmed,Waheed Ahmed Abro

DOI: https://doi.org/10.1109/ICCWAMTIP47768.2019.9067714

2019-01-01

Abstract:Low- light Image enhancement is a fundamental problem in computer vision. Over the years, many efforts have been invested in improving the quality of images captured under low- lightning conditions. However, the issue of low-light image enhancement is still infancy into its experimental findings. Therefore, to meet this objective, in this paper, a method to low-light image enhancement based on wavelet transform is proposed. First, the RGB input image is converted into HSV color space to obtain the V channel of the image. Second, the V-channel is decomposed to get the concentrated information in its low and high-frequency components using dual-tree complex wavelet transform. Since the illumination is concentrated on the low- frequency component of the image, an adaptive illumination boost algorithm is applied to enhance the illumination of the image. Besides, edges are preserved on the high-frequency component of the image, anisotropic diffusion is applied to denoise the image. Finally, the image is reconstructed by using an inverse dual-tree complex wavelet transform and converted back into RGB space to achieve improved outcomes. Experimental results performed to justify the effectiveness and robustness of the proposed method with an average Entropy of (7.577555), NIQE of (19.053795), VIF of (5.030787) and VLD of (4.1350945).