Night Image Enhancement
Under extreme low-lighting conditions, images have low contrast, low brightness, and high noise. In this paper, we propose a principal component analysis framework to enhance low-light-level images with decomposed luminance-chrominance components. A multi-scale retinex-based adaptive filter is developed for the luminance component to enhance contrast and brightness significantly. Noise is attenuated by a proposed collaborative filtering employed to both the luminance and chrominance components that reveal every finest detail by preserving the unique features in the image. To evaluate the effectiveness of the proposed algorithm, a simulation model is proposed to generate nighttime images for various levels of contrast and noise. The proposed algorithm can process a wide range of images without introducing ghosting and halo artifacts. The quantitative performance of the algorithm is measured in terms of both full-reference and blind performance metrics. It shows that the proposed method delivers state-of-the-art performance both in terms of objective criteria and visual quality compared to the existing methods.
Algorithm
Framework of our Algorithm
Experimental results
Experiments were conducted with real night images under various scenes for different lighting conditions with and without human objects for subjective and objective evaluations. The proposed algorithm is compared with other existing contrast enhancement and denoising algorithms for various levels of contrast, noise, and their results are compared. Images were captured using DSLR camera and the resolution of the testing images is 1440 X 1080.
Enhancement results of the proposed algorithm
(a)-(d) Real night images of different scenes, (e)-(h) Corresponding results.
Comparative analysis of different algorithms with real night images
Average results of various real night images for different evaluation metric
Demo Video
Resources
Publication
1. Steffi Agino Priyanka, Yuan-Kai Wang, and Shih-Yu Huang, "Low-Light Image Enhancement by Principal Component Analysis," IEEE Access, Vol.7(1), pp. 3082-3092, Dec.2018. [PDF]
2. Steffi Agino Priyanka, Hsiao-Jung Tung, Yuan-Kai Wang, "Contrast Enhancement of Night Images", 16th IEEE conference on Machine learning and Cybernetics, PP 380-385, July 2016. [PDF]
2. Steffi Agino Priyanka, Hsiao-Jung Tung, Yuan-Kai Wang, "Contrast Enhancement of Night Images", 16th IEEE conference on Machine learning and Cybernetics, PP 380-385, July 2016. [PDF]