Conditional Laplacian pyramid networks for exposure correction

Mengyuan Huang, Kan Chang, Qingpao Qin, Yahui Tang, Guiqing Li

Abstract: Highlights•The proposed CLPN can correct various exposure errors in the same framework.•Components decomposed by Laplacian pyramid are enhanced in a coarse-to-fine manner.•Conditional feature is extracted and used to guide the correction of LF features.•Correlations between LF and HF components are explored by affine transformation.•The proposed method outperforms many state-of-the-art methods on various datasets.