NTIRE 2022 Burst Super-Resolution Challenge

Bhat, Goutam; Danelljan, Martin; Timofte, Radu; Cao, Yizhen; Cao, Yuntian; Chen, Meiya; Chen, Xihao; Shen, Cheng,; Dudhane, Akshay; Fan, Haoqiang; Gang, Ruipeng; Gao, Jianliang; Gu, Yu; Huang, Jie; Huang, Liufeng; Jo, Youngsu; Kang, Suk‐Ju; Khan, Salman; Khan, Fahad Shahbaz; Kondo, Yuki; Li, Chenghua; Li, Fangya; Li, Jinjing; Li, Youwei; Li, Zechao; Liu, Chenming; Liu, Shuaicheng; Liu, Zikun; Liu, Zhuoming; Luo, Ziwei; Luo, Zhaohui; Mehta, Nancy; Murala, Subrahmanyam; Nam, Yoonchan; Nakatani, Chihiro; Ostyakov, Pavel; Pan, Jinshan; Song, Ge; Sun, Jian; Sun, Lei; Tang, Jinhui; Ukita, Norimichi; Wen, Zhihong; Wu, Qi; Wu, Xiaohe; Xiao, Zeyu; Xiong, Zhiwei; Xu, Rongjian; Xu, Ruikang; Yan, Youliang; Yang, Jialin; Yang, Wentao; Yang, Zhong‐Bao; Yasue, Fuma; Yao, Mingde; Yu, Lei; Zhang, Cong; Zamir, Syed Waqas; Zhang, Jianxing; Zhang, Shuohao; Zhang, Zhilu; Zheng, Qian; Zhou, Guoqing; Zhussip, Magauiya; Zou, Xueyi; Zuo, Wangmeng

doi:10.1109/cvprw56347.2022.00117

Cited by 19 publications

(6 citation statements)

References 41 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The image reconstruction that is effective for our diffractive optic-based imaging system is similar to a single image super-resolution (SISR) task. There are various deep-learning solutions that can produce visual-pleasing results with high PSNR and SSIM values for a SISR task [ 26 , 27 ]. Most of these methods are based on the known image degradation models and range from simple downsampling with a bicubic upsampling [ 28 , 29 , 30 , 31 , 32 ] to more recent works, relying on blurring kernel degradation [ 30 , 31 , 32 ].…”

Section: Deep Learning-based Image Reconstructionmentioning

confidence: 99%

See 1 more Smart Citation

Hybrid Refractive-Diffractive Lens with Reduced Chromatic and Geometric Aberrations and Learned Image Reconstruction

Evdokimova

Подлипнов

Ivliev

et al. 2022

Sensors

View full text Add to dashboard Cite

In this paper, we present a hybrid refractive-diffractive lens that, when paired with a deep neural network-based image reconstruction, produces high-quality, real-world images with minimal artifacts, reaching a PSNR of 28 dB on the test set. Our diffractive element compensates for the off-axis aberrations of a single refractive element and has reduced chromatic aberrations across the visible light spectrum. We also describe our training set augmentation and novel quality criteria called “false edge level” (FEL), which validates that the neural network produces visually appealing images without artifacts under a wide range of ISO and exposure settings. Our quality criteria (FEL) enabled us to include real scene images without a corresponding ground truth in the training process.

show abstract

Section: Deep Learning-based Image Reconstructionmentioning

confidence: 99%

“…We use a modification of the U-Net architecture [ 26 ], which was successfully applied for post-processing images captured by harmonic diffractive lenses [ 27 , 28 ]. The original U-Net architecture as follows:…”

Section: Deep Learning-based Image Reconstructionmentioning

confidence: 99%

Hybrid Refractive-Diffractive Lens with Reduced Chromatic and Geometric Aberrations and Learned Image Reconstruction

Evdokimova

Подлипнов

Ivliev

et al. 2022

Sensors

View full text Add to dashboard Cite

show abstract

“…Multi-frame super-resolution can additionally leverage inter-frame complementary information, recovering more details compared to single-frame super-resolution, and has broad application prospects in fields such as computational photography [26,27], remote sensing satellite imaging [28,29], etc. Commonly used MFSR datasets include PROBA-V [30] for remote sensing tasks and BurstSR [31] for computational photography tasks. The performance improvement of multi-frame superresolution relies on the sub-pixel-level alignment of multiple frames, and the accuracy of alignment directly impacts the reconstruction results.…”

Section: Introductionmentioning

confidence: 99%

Burst-Enhanced Super-Resolution Network (BESR)

Li,

Lv,

Zhang

et al. 2024

Sensors

View full text Add to dashboard Cite

Multi-frame super-resolution (MFSR) leverages complementary information between image sequences of the same scene to increase the resolution of the reconstructed image. As a branch of MFSR, burst super-resolution aims to restore image details by leveraging the complementary information between noisy sequences. In this paper, we propose an efficient burst-enhanced super-resolution network (BESR). Specifically, we introduce Geformer, a gate-enhanced transformer, and construct an enhanced CNN-Transformer block (ECTB) by combining convolutions to enhance local perception. ECTB efficiently aggregates intra-frame context and inter-frame correlation information, yielding an enhanced feature representation. Additionally, we leverage reference features to facilitate inter-frame communication, enhancing spatiotemporal coherence among multiple frames. To address the critical processes of inter-frame alignment and feature fusion, we propose optimized pyramid alignment (OPA) and hybrid feature fusion (HFF) modules to capture and utilize complementary information between multiple frames to recover more high-frequency details. Extensive experiments demonstrate that, compared to state-of-the-art methods, BESR achieves higher efficiency and competitively superior reconstruction results. On the synthetic dataset and real-world dataset of BurstSR, our BESR achieves PSNR values of 42.79 dB and 48.86 dB, respectively, outperforming other MFSR models significantly.

show abstract

“…This challenge is one of the NTIRE 2022 associated challenges: spectral recovery [7], spectral demosaicing [6], perceptual image quality assessment [19], inpainting [40], night photography rendering [17], efficient super-resolution [31], learning the super-resolution space [34], super-resolution and quality enhancement of compressed video [49], high dynamic range [38], stereo super-resolution [45], burst super-resolution [8].…”

Section: Introductionmentioning

confidence: 99%

NTIRE 2022 Spectral Recovery Challenge and Data Set

Arad¹,

Timofte²,

Yahel³

et al. 2022

2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)

Self Cite

View full text Add to dashboard Cite

This paper reviews the third biennial challenge on spectral reconstruction from RGB images, i.e., the recovery of whole-scene hyperspectral (HS) information from a 3-channel RGB image. This challenge presents the "ARAD 1K" data set: a new, larger-than-ever natural hyperspectral image data set containing 1,000 images. Challenge participants were required to recover hyperspectral information from synthetically generated JPEGcompressed RGB images simulating capture by a known calibrated camera, operating under partially known parameters, in a setting which includes acquisition noise. The challenge was attended by 241 teams, with 60 teams competing in the final testing phase, 12 of which provided detailed descriptions of their methodology which are included in this report. The performance of these submissions is reviewed and provided here as a gauge for the current stateof-the-art in spectral reconstruction from natural RGB images.

show abstract

NTIRE 2022 Burst Super-Resolution Challenge

Cited by 19 publications

References 41 publications

Hybrid Refractive-Diffractive Lens with Reduced Chromatic and Geometric Aberrations and Learned Image Reconstruction

Hybrid Refractive-Diffractive Lens with Reduced Chromatic and Geometric Aberrations and Learned Image Reconstruction

Burst-Enhanced Super-Resolution Network (BESR)

NTIRE 2022 Spectral Recovery Challenge and Data Set

Contact Info

Product

Resources

About