End-to-End Exposure Fusion Using Convolutional Neural Network

Wang, Jinhua; Wang, Weiqiang; Xu, Guili; Liu, Hongzhe

doi:10.1587/transinf.2017edl8173

Cited by 18 publications

(15 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Since then, many MEF algorithms based on deep learning have been proposed. In 2018, Wang [75] proposed a supervised CNN-based framework for MEF. The main innovation of the approach was that it used the CNN model to gain multiple sub-images of the input images to use more neighborhood information for convolution operation.…”

Section: Supervised Methodsmentioning

confidence: 99%

Multi-Exposure Image Fusion Techniques: A Comprehensive Review

Liu

Song

et al. 2022

Remote Sensing

View full text Add to dashboard Cite

Multi-exposure image fusion (MEF) is emerging as a research hotspot in the fields of image processing and computer vision, which can integrate images with multiple exposure levels into a full exposure image of high quality. It is an economical and effective way to improve the dynamic range of the imaging system and has broad application prospects. In recent years, with the further development of image representation theories such as multi-scale analysis and deep learning, significant progress has been achieved in this field. This paper comprehensively investigates the current research status of MEF methods. The relevant theories and key technologies for constructing MEF models are analyzed and categorized. The representative MEF methods in each category are introduced and summarized. Then, based on the multi-exposure image sequences in static and dynamic scenes, we present a comparative study for 18 representative MEF approaches using nine commonly used objective fusion metrics. Finally, the key issues of current MEF research are discussed, and a development trend for future research is put forward.

show abstract

Section: Supervised Methodsmentioning

confidence: 99%

Multi-Exposure Image Fusion Techniques: A Comprehensive Review

Liu

Song

et al. 2022

Remote Sensing

View full text Add to dashboard Cite

show abstract

“…Many of them attempt to do this using a single shot 21,22 or by simply having a extreme number of weights. [21][22][23][24] For real-time, robust driver recognition, both of these are unacceptable. In the former case, we expect that a single shot will have a high probability of either misfiring or inaccurately capturing the face, due to the potential for obstructive glare.…”

Section: Related Workmentioning

confidence: 99%

The Mertens Unrolled Network (MU-Net): A high dynamic range fusion neural network for through the windshield driver recognition

Ruby

Bolme

Brogan

et al. 2020

Autonomous Systems: Sensors, Processing, and Security for Vehicles and Infrastructure 2020

View full text Add to dashboard Cite

Face recognition of vehicle occupants through windshields in unconstrained environments poses a number of unique challenges ranging from glare, poor illumination, driver pose and motion blur. In this paper, we further develop the hardware and software components of a custom vehicle imaging system to better overcome these challenges. After the build out of a physical prototype system that performs High Dynamic Range (HDR) imaging, we collect a small dataset of through-windshield image captures of known drivers. We then re-formulate the classical Mertens-Kautz-Van Reeth HDR fusion algorithm as a pre-initialized neural network, which we name the Mertens Unrolled Network (MU-Net), for the purpose of fine-tuning the HDR output of through-windshield images. Reconstructed faces from this novel HDR method are then evaluated and compared against other traditional and experimental HDR methods in a pre-trained state-of-the-art (SOTA) facial recognition pipeline, verifying the efficacy of our approach.

show abstract

“…In these methods, two source images with different exposures are directly input into a fusion network, and the fused image is obtained from the output of the network. The fusion networks can be trained in a common supervised way using ground truth fusion images (Zhang et al 2020b;Wang et al 2018;Li and Zhang 2018) or in an unsupervised way by encouraging the fused image to retain different aspects of the important information in the source images (Xu et al 2020a;Ram Prabhakar, Sai Srikar, and Venkatesh Babu 2017;Xu et al 2020b;Zhang et al 2020a;Ma et al 2019b). However, both supervised and unsupervised MEF methods require a large amount of multi-exposure data for training.…”

Section: Introductionmentioning

confidence: 99%

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

Qu¹,

Liu²,

Wang³

et al. 2021

Preprint

View full text Add to dashboard Cite

In this paper, we propose TransMEF, a transformer-based multi-exposure image fusion framework that uses selfsupervised multi-task learning. The framework is based on an encoder-decoder network, which can be trained on large natural image datasets and does not require ground truth fusion images. We design three self-supervised reconstruction tasks according to the characteristics of multi-exposure images and conduct these tasks simultaneously using multi-task learning; through this process, the network can learn the characteristics of multi-exposure images and extract more generalized features. In addition, to compensate for the defect in establishing long-range dependencies in CNN-based architectures, we design an encoder that combines a CNN module with a transformer module. This combination enables the network to focus on both local and global information. We evaluated our method and compared it to 11 competitive traditional and deep learning-based methods on the latest released multi-exposure image fusion benchmark dataset, and our method achieved the best performance in both subjective and objective evaluations. Code will be available at https://github.com/miccaiif/TransMEF.

show abstract

End-to-End Exposure Fusion Using Convolutional Neural Network

Cited by 18 publications

References 10 publications

Multi-Exposure Image Fusion Techniques: A Comprehensive Review

Multi-Exposure Image Fusion Techniques: A Comprehensive Review

The Mertens Unrolled Network (MU-Net): A high dynamic range fusion neural network for through the windshield driver recognition

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

Contact Info

Product

Resources

About