Perceptual Quality Assessment for Multi-Exposure Image Fusion

By Kede Ma, Kai Zeng, Zhou Wang. IEEE Transactions on Image Processing, 2015

1,117 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Media Technology

View Paper: https://www.semanticscholar.org/paper/19986f7af0086f0e329779f3497e97440765d170

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

1. Luc Van Gool
2. Chen Change Loy
3. Lei Zhang
4. Radu Timofte
5. Jiayi Ma
6. Ming-Hsuan Yang
7. Dacheng Tao
8. Ling Shao

In Media Technology

1. Jiayi Ma
2. Hui Li
3. Xiaojun Wu
4. Radu Timofte
5. Lei Zhang
6. Yulun Zhang
7. Luc Van Gool
8. Chen Change Loy

Citations in the corpus

• Learning a Simple Low-Light Image Enhancer from Paired Low-Light Instances (2023, methodology): “Furthermore, we adopt the MEF dataset [21] for visual comparisons.”
• EnlightenGAN: Deep Light Enhancement Without Paired Supervision (2019, methodology): “es of medium brightness. All these photos are converted to PNG format and resized to 600 400 pixels. For testing images, we choose those standard ones used in previous works (NPE [19], LIME [21], MEF [44], DICM [45], VV, 2 etc.). EnlightenGAN is first trained from the scratch for 100 epochs with the ”
• Blind Image Quality Assessment Using a Deep Bilinear Convolutional Neural Network (2019, methodology): “Following [19], we synthesize distorted images with five degradation levels except for over-exposure and underexposure, for which only two levels are generated [32].”
• Infrared and Visible Image Fusion with ResNet and zero-phase component analysis (2018, background): “These metrics include: En(entropy), MI(mutual information), Q abf [34], F MI pixel [32], F MI dct [32], F MI w [32], N abf [33], SCD[35], SSIM a [17], and MS SSIM[36].”
• GLADNet: Low-Light Enhancement Network with Global Awareness (2018, methodology): “We evaluate our approach on real under-exposed images from public LIME-data [4], DICM [14], and MEF [15]”
• R2RNet: Low-light Image Enhancement via Real-low to Real-normal Network (2021, methodology): “We compare the proposed method with the existing state-of-art methods (MF, Dong, NPE, SRIE, BIMEF, MSRCR, LIME, RetinexNet, DSLR, MBLLEN, EnlightenGAN, and Zero-DCE) on six publicly available datasets, including LOL, LIME, DICM [50], NPE [51], MEF [52], and VV 1 .”
• Retinexformer: One-stage Retinex-based Transformer for Low-light Image Enhancement (2023, methodology): “In addition to the above eight benchmarks, we test our method on five datasets: LIME [18], NPE [50], MEF [36], DICM [28], and VV [47] that have no ground truth.”
• Attention Guided Low-Light Image Enhancement with a Large Scale Low-Light Simulation Dataset (2019, background): “Dynamic Range (HDR) dataset, like MEF dataset (Ma et al. 2015), for testing.”
• Rethinking the Image Fusion: A Fast Unified Image Fusion Network based on Proportional Maintenance of Gradient and Intensity (2020, background): “In multi-exposure image fusion, Prabhakar et al. (Ram Prabhakar, Sai Srikar, and Venkatesh Babu 2017) proposed an unsupervised deep learning framework that utilizes a no-reference quality metric MEF-SSIM (Ma, Zeng, and Wang 2015) as the loss function and can produce satisfactory fused results.”
• Zero-Reference Deep Curve Estimation for Low-Light Image Enhancement (2020, methodology): “We process low-light images from the image sets (NPE, LIME, MEF, DICM, VV) by different methods.”
• Sparse Gradient Regularized Deep Retinex Network for Robust Low-Light Image Enhancement (2021, methodology): “Compared with the datasets generated by Multi-Exposure image Fusion (MEF) [42] or High Dynamic Range (HDR) algorithm [16], our real captured paired data is more representative and can provide more useful guidance to train a low-light image enhancer.”
• DeepFuse: A Deep Unsupervised Approach for Exposure Fusion with Extreme Exposure Image Pairs (2017, methodology): “[15], Mertens and GFF results are ranked better than other MEF algorithms”
• Deep Retinex Decomposition for Low-Light Enhancement (2018, methodology): “Several low-light enhancement works use datasets of High Dynamic Range (HDR) as an alternative, such as MEF dataset [18].”
• DenseFuse: A Fusion Approach to Infrared and Visible Images (2018, methodology): “Methods En Qabf [28] SCD [29] FMIw [30] FMIdct [30] SSIMa MS SSIM [31] CBF [25] 6.”
• Learning to Enhance Low-Light Image via Zero-Reference Deep Curve Estimation (2021, neutral): “NPE [17] (84 images), LIME [19] (10 images), MEF [40] (17 images), DICM [41] (64 images), and VV(1) (24 images).”
• MDLatLRR: A Novel Decomposition Method for Infrared and Visible Image Fusion (2018, background): “…[1] calculates image quality metric value based on the sum of the correlations of differences; a modified structural similarity SSIM a [21]; MS-SSIM [34] calculates a modified structural similarity which only focuses on structural information; visual information fidelity( V IF ) [40]; and edge…”
• Robust Multi-Exposure Image Fusion: A Structural Patch Decomposition Approach (2017, methodology): “MEF) that well correlates with subjective judgments [7].”