Abstract:Recently, learning-based multi-exposure fusion (MEF) methods have made significant improvements. However, these methods mainly focus on static scenes and are prone to generate ghosting artifacts when tackling a more common scenario, i.e., the input images include motion, due to the lack of a benchmark dataset and solution for dynamic scenes. In this paper, we fill this gap by creating an MEF dataset of dynamic scenes, which contains multi-exposure image sequences and their corresponding high-quality reference images. To construct such a dataset, we propose a 'static-for-dynamic' strategy to obtain multi-exposure sequences with motions and their corresponding reference images. To the best of our knowledge, this is the first MEF dataset of dynamic scenes. Correspondingly, we propose a deep dynamic MEF (DDMEF) framework to reconstruct a ghost-free high-quality image from only two differently exposed images of a dynamic scene. DDMEF is achieved through two steps: pre-enhancement-based alignment and privilege-information-guided fusion. The former pre-enhances the input images before alignment, which helps to address the misalignments caused by the significant exposure difference. The latter introduces a privilege distillation scheme with an information attention transfer loss, which effectively improves the deghosting ability of the fusion network. Extensive qualitative and quantitative experimental results show that the proposed method outperforms state-of-the-art dynamic MEF methods. The source code and dataset are released at https://github.com/Tx000/Deep_dynamicMEF.

Deep Guided Learning for Fast Multi-Exposure Image Fusion

Deep Guided Learning for Fast Multi-Exposure Image Fusion

Efficient Multi-exposure Image Fusion Via Filter-dominated Fusion and Gradient-driven Unsupervised Learning.

Ref-MEF: Reference-Guided Flexible Gated Image Reconstruction Network for Multi-Exposure Image Fusion

Multi-exposure image fusion via deep perceptual enhancement

StackMFF: End-to-end Multi-Focus Image Stack Fusion Network

Fast detail-preserving exposure fusion

MEFLUT: Unsupervised 1D Lookup Tables for Multi-exposure Image Fusion

Multi Exposure Fusion for High Dynamic Range Imaging Via Multi-Channel Gradient Tensor

MEFNet: Multi-scale Event Fusion Network for Motion Deblurring

Multi-Exposure Image Fusion via Multi-Scale and Context-Aware Feature Learning

Deep Multi-Exposure Image Fusion for Dynamic Scenes

EMEF: Ensemble Multi-Exposure Image Fusion

MobileMEF: Fast and Efficient Method for Multi-Exposure Fusion

TransMEF: A Transformer-Based Multi-Exposure Image Fusion Framework Using Self-Supervised Multi-Task Learning

CurveMEF: Multi-eposure fusion via curve embedding network

Multi-Scale Exposure Fusion Via Content Adaptive Edge-Preserving Smoothing Pyramids.

MobileMEF: fast and efficient method for real-time mobile multi-exposure fusion

Rethinking Multi-Exposure Image Fusion with Extreme and Diverse Exposure Levels: A Robust Framework Based on Fourier Transform and Contrastive Learning.

Multi-Exposure Image Fusion via Deformable Self-Attention

Multiexposure Fusion Based on Multiscale Edge Information Measurement and Adaptive Pyramid Optimization