AAAI2020论文,低光照增强论文。
论文使用多曝光融合先产生first-stage的结果,然后进行edge information refinement进行refine。提出的EEMEFM旨在recover well-exposed image details, reduce noise and color bias, and maintain sharp edges for extremely low-light image enhancement.
网络中使用了一个fusion模块,还是很不错的,具有参考意义