Low Illumination Image Enhancement Based on Gaussian Fusion Strategy
Yiwen Dou, Hong-chao Liao, Xia Chai, Liping Zhang, Zi-ang Chen · 2021
Under low illumination environment, the image captured by ordinary visual capture equipment has fuzzy details resulting in low recognition rate. Meanwhile, traditional enhancement algorithm has defects in noise amplification, color distortion and low contrast. Based on multiple exposure fusion principle and virtual exposure idea, this paper proposes an image enhancement algorithm based on Gaussian fusion strategy. The proposed algorithm first generates pseudo-multiple exposure maps sequences combined with a variety of emerging enhancement algorithms, where the adopted enhancement algorithms include camera response model, adaptive Sigmoid transfer function and homomorphic filtering enhancements. Subsequently, the Gaussian weighted fusion of the obtained sequence of pseudo-multiple exposure maps is performed. Finally, the resulting fusion image performs the contrast enhancement of the screened Poisson equation with the final enhancement result. The experimental results show that the algorithm enhancement results have superior data of objective image quality evaluation index which shows the effectiveness of the algorithm in image enhancement.