TY - JOUR
T1 - A degradation model for simultaneous brightness and sharpness enhancement of low-light image
AU - Li, Pengliang
AU - Liang, Junli
AU - Zhang, Miaohua
N1 - Publisher Copyright:
© 2021
PY - 2021/12
Y1 - 2021/12
N2 - Although a large number of methods have been proposed for low-light image enhancement, there still remain challenges for these methods to simultaneously achieve excellent sharpness/resolution, high calculation efficiency as well as visual pleasure requirements. In this communication, we propose a new low-light image enhancement method based on the degradation model to overcome this dilemma. Specifically, we regard the low-light image enhancement as a special inverse problem of image degradation, and then the task of low-light enhancement is logically embedded in the iterative back-projection (IBP) framework. Meanwhile, an adaptive gamma correction is utilized to adaptively adjust the brightness, and then the IBP framework is transferred to the logarithmic domain instead of the spatial domain for further acceleration. Besides, a simple and effective pre-processing strategy is proposed to pre-enhance the low-light input while making the enhanced image clarify (or visual pleasure). Extensive experimental results on public databases and seven state-of-the-art benchmarks consistently demonstrate the effectiveness and efficiency of the proposed method both visually and quantitatively.
AB - Although a large number of methods have been proposed for low-light image enhancement, there still remain challenges for these methods to simultaneously achieve excellent sharpness/resolution, high calculation efficiency as well as visual pleasure requirements. In this communication, we propose a new low-light image enhancement method based on the degradation model to overcome this dilemma. Specifically, we regard the low-light image enhancement as a special inverse problem of image degradation, and then the task of low-light enhancement is logically embedded in the iterative back-projection (IBP) framework. Meanwhile, an adaptive gamma correction is utilized to adaptively adjust the brightness, and then the IBP framework is transferred to the logarithmic domain instead of the spatial domain for further acceleration. Besides, a simple and effective pre-processing strategy is proposed to pre-enhance the low-light input while making the enhanced image clarify (or visual pleasure). Extensive experimental results on public databases and seven state-of-the-art benchmarks consistently demonstrate the effectiveness and efficiency of the proposed method both visually and quantitatively.
KW - Adaptive gamma correction
KW - Iterative back-projection
KW - Logarithmic domain acceleration
KW - Low-light image enhancement
UR - http://www.scopus.com/inward/record.url?scp=85113832942&partnerID=8YFLogxK
U2 - 10.1016/j.sigpro.2021.108298
DO - 10.1016/j.sigpro.2021.108298
M3 - 文章
AN - SCOPUS:85113832942
SN - 0165-1684
VL - 189
JO - Signal Processing
JF - Signal Processing
M1 - 108298
ER -