Learning Exposure Correction in Dynamic Scenes

TL;DR

This paper constructs the first real-world paired video dataset, including both underexposure and overexposure dynamic scenes, and proposes an end-to-end video exposure correction network, in which a dual-stream module is designed to deal with both underexposure and overexposure factors.

Abstract

Exposure correction aims to enhance visual data suffering from improper exposures, which can greatly improve satisfactory visual effects. However, previous methods mainly focus on the image modality, and the video counterpart is less explored in the literature. Directly applying prior image-based methods to videos results in temporal incoherence with low visual quality. Through thorough investigation, we find that the development of relevant communities is limited by the absence of a benchmark dataset. Therefore, in this paper, we construct the first real-world paired video dataset, including both underexposure and overexposure dynamic scenes. To achieve spatial alignment, we utilize two DSLR cameras and a beam splitter to simultaneously capture improper and normal exposure videos. Additionally, we propose an end-to-end video exposure correction network, in which a dual-stream module is designed to deal with both underexposure and overexposure factors, enhancing the illumination based on Retinex theory. The extensive experiments based on various metrics and user studies demonstrate the significance of our dataset and the effectiveness of our method. The code and dataset are available at https://github.com/kravrolens/VECNet.

Figures (17)

• Figure 1: The above sub-figure is the benchmark for our proposed video exposure correction, while the below sub-figures are the visual comparison between an image-based method LACT baek2023luminance and our proposed method. LACT takes single frames as input and results in temporal exposure incoherence with low visual quality. Our method utilizes temporal information to achieve consecutive exposures.
• Figure 2: We built the optical system to capture the normal and over-/under- exposure video pairs.
• Figure 3: Statistics on our DIME dataset. (a) The input-ground truth brightness mapping curve statistics. (b) Luminance distribution for overexposure videos. (c) Luminance distribution for underexposure videos. (d) Luminance distribution for ground truth videos.
• Figure 4: LOE and optical flow of different datasets.
• Figure 5: Example videos from our DIME dataset cover under-/over-exposures, indoor/outdoor, and camera/object motions.