본문 바로가기

Learning Lightweight Low-Light Enhancement Network Using Pseudo Well-Exposed Images

문서 유형
연구출판물 · 학술지 논문
등재
SCIEScopus
저널 순위
JCR Q2Scopus Q1
저널
IEEE SIGNAL PROCESSING LETTERS29 · pp.289 · 2022
피인용
Web of Science 17Scopus 20
FWCI (Scopus)
1.39
JCR IF
JCR 2021 3.201
ISSN
1070-9908

저자

  1. Ko, Seonggwan제1저자Chungnam Natl Univ, Dept Comp Sci & Engn, Daejeon 34134, South Korea
  2. 박지선Pusan Natl Univ, Sch Comp Sci & Engn, Busan 46241, South Korea
  3. Chae, ByungjooChungnam Natl Univ, Dept Elect Engn, Daejeon 34134, South Korea
  4. Cho, Donghyeon교신저자Chungnam Natl Univ, Dept Elect Engn, Daejeon 34134, South Korea

초록

Recently, there has been growing attention on deep learning-based low-light image enhancement algorithms. With this interest, various synthetic low-light image datasets have been released publicly. However, real-world low-light and well-exposed image pair datasets are still lacking. In this paper, we propose a real-world low-light image dataset and a practical lightweight low-light image enhancement network. In order to construct a large-scale real-world low-light dataset, we have not only captured under-exposed images by ourselves but also collected under-exposed images from the Internet. Then, we produce pseudo well-exposed images for each low-light image. Using pairs of a real-world low-light image and a pseudo well-exposed image, we present a lightweight deep CNN model through knowledge distillation. Experimental results demonstrate the effectiveness and practicality of the proposed method on various datasets.

키워드

TrainingFeature extractionKnowledge engineeringImage enhancementLightingDynamic rangeComputational modelingLow-light enhancementpseudo labelsknowledge distillationDEEP CNNRETINEX

식별자

Scopus EID
2-s2.0-85124214801
WoS UID
000748371000009