Article
Computer Science, Artificial Intelligence
Qiucheng Wu, Yifan Jiang, Junru Wu, Victor Kulikov, Vidit Goel, Nikita Orlov, Humphrey Shi, Zhangyang Wang, Shiyu Chang
Summary: In this work, a super-network called Ada-Deblur is proposed, which can be applied to a wide range of blur levels without re-training. By dynamically adapting network architectures from a well-trained super-network structure, it achieves flexible image processing with different deblurring capacities at test time. Extensive experiments demonstrate that this method outperforms strong baselines in terms of reconstruction accuracy while incurring minimal computational overhead. Additionally, it is effective for both synthetic and realistic blurs, and performs particularly well on unseen and strong blur levels.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2023)
Article
Computer Science, Artificial Intelligence
Sanghyun Son, Jaeha Kim, Wei-Sheng Lai, Ming-Hsuan Yang, Kyoung Mu Lee
Summary: This study proposes a novel method to simulate an unknown downsampling process without imposing restrictive prior knowledge. By designing a low-frequency loss (LFL) and an adaptive data loss (ADL) in the adversarial training framework, the downsampling model becomes more generalizable and enables more accurate reconstructions.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2022)
Article
Computer Science, Information Systems
Sujy Han, Tae Bok Lee, Yong Seok Heo
Summary: This paper introduces a noise-robust and stable framework based on deep image prior for single image super-resolution tasks. By utilizing generative adversarial networks and self-supervision loss for noise estimation, the proposed method outperforms existing approaches for noisy images.
Article
Computer Science, Information Systems
Xin Deng, Hao Wang, Mai Xu, Li Li, Zulin Wang
Summary: This paper proposes a novel latitude-aware upscaling network, LAU-Net+, for omnidirectional image super-resolution (ODI-SR) task. The network fully considers the characteristics of ODIs and saves computational resources by learning different upscaling factors for different latitude bands. Experimental results demonstrate that LAU-Net+ achieves state-of-the-art results on various ODI datasets.
IEEE TRANSACTIONS ON MULTIMEDIA
(2023)
Article
Computer Science, Artificial Intelligence
Li Ji, Qinghui Zhu, Yongqin Zhang, Juanjuan Yin, Ruyi Wei, Jinsheng Xiao, Deqiang Xiao, Guoying Zhao
Summary: This paper proposes a novel high-performance cross-domain heterogeneous residual network for super resolved image reconstruction. The network models heterogeneous residuals between different feature layers through hierarchical residual learning. Experimental results demonstrate that the proposed method achieves state-of-the-art performance on multiple datasets.
Article
Computer Science, Artificial Intelligence
Cheng Ma, Yongming Rao, Jiwen Lu, Jie Zhou
Summary: Structures play a crucial role in single image super-resolution (SISR), and this paper proposes a structure-preserving super-resolution (SPSR) method to address the issue of structural distortions. By utilizing gradient guidance and a learnable neural structure extractor (NSE), our method achieves superior results in both detail recovery and structure preservation.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2022)
Article
Computer Science, Artificial Intelligence
Debabrata Pal, Shirsha Bose, Deeptej More, Ankit Jha, Biplab Banerjee, Yogananda Jeppu
Summary: In this paper, we propose a Multi-scale Optimized Attention-aware Meta-Learning framework for SR (MAML-SR), which explores the multi-scale hierarchical self-similarity of recurring patches in a test image. Without any pre-training, the model is directly meta-trained with a second-order optimization, using first-order adapted parameters from intermediate scales. Additionally, a novel cross-scale spectro-spatial attention learning unit is used to maximize non-local self-similarity and amplify salient edges.
PATTERN RECOGNITION LETTERS
(2023)
Article
Computer Science, Information Systems
Jiun Lee, Inyong Yun, Jaekwang Kim
Summary: The EPSR method uses the mFRN structure to reconstruct high-resolution images, preserving structural information and restoring texture. Experimental results demonstrate that EPSR performs competitively in terms of PSNR, SSIM evaluation metrics, and visual results.
Article
Computer Science, Artificial Intelligence
Xinyi Ren, Qiang Hui, Xingke Zhao, Jianping Xiong, Jun Yin
Summary: This study proposes a fidelity-controllable face super-resolution (FSR) network called BESRGAN, which introduces a fidelity ratio to control the adversarial effect on the generator. It also designs an equilibrium perceptual discriminator to match perception loss distributions and guides the generator to produce more realistic faces. Furthermore, a channel-spatial attention module (CSAM) is proposed to eliminate local distortions and improve visual performance.
IET IMAGE PROCESSING
(2023)
Article
Computer Science, Hardware & Architecture
Zhijie Tang, Siyu Yan, Congqi Xu
Summary: This paper introduces an adaptive super-resolution image reconstruction algorithm based on fractal theory. It classifies the texture complexity of the image and improves the reconstruction quality using fractal geometry and rational interpolation. It also proposes an image segmentation algorithm based on local fractal dimension, and the experimental results show satisfactory performance. The paper lays the theoretical foundation for the application of fractals in super-resolution image reconstruction.
Article
Computer Science, Information Systems
Yan Wang, Tongtong Su, Yusen Li, Jiuwen Cao, Gang Wang, Xiaoguang Liu
Summary: This paper proposes a lightweight network called DDistill-SR, which significantly improves the quality of super-resolution by capturing and reusing more helpful information. By using plug-in reparameterized dynamic units (RDU) and dynamic distillation fusion (DDF) modules, the network is able to achieve better performance while reducing parameters and computational overhead.
IEEE TRANSACTIONS ON MULTIMEDIA
(2023)
Article
Computer Science, Artificial Intelligence
Fayaz Ali Dharejo, Iyyakutti Iyappan Ganapathi, Muhammad Zawish, Basit Alawode, Moath Alathbah, Naoufel Werghi, Sajid Javed
Summary: The resource-limited nature of underwater vision equipment affects underwater robotics and ocean engineering tasks. Super-resolution methods, particularly using Vision Transformers (ViTs), have emerged to enhance low-resolution underwater images. However, ViTs face challenges in handling severe degradation in underwater imaging. In contrast, Multi-scale ViTs (MViTs) overcome these challenges by preserving long-range dependencies through evolving channel capacity. This study proposes a novel algorithm, SwinWave-SR, for efficient and accurate multi-scale super-resolution for underwater images.
INFORMATION FUSION
(2024)
Article
Computer Science, Information Systems
Ke-Jia Chen, Mingyu Wu, Yibo Zhang, Zhiwei Chen
Summary: This paper proposes a super-resolution network SR-AFU based on adaptive frequency component upsampling, which achieves faster training speed and more realistic visual effects through the combination of multiple residual blocks and convolutional upsampling blocks.
FRONTIERS OF COMPUTER SCIENCE
(2023)
Article
Biochemical Research Methods
Linjing Fang, Fred Monroe, Sammy Weiser Novak, Lyndsey Kirk, Cara R. Schiavon, Seungyoon B. Yu, Tong Zhang, Melissa Wu, Kyle Kastner, Alaa Abdel Latif, Zijun Lin, Andrew Shaw, Yoshiyuki Kubota, John Mendenhall, Zhao Zhang, Gulcin Pekkurnaz, Kristen Harris, Jeremy Howard, Uri Manor
Summary: Point-scanning imaging systems are widely used for high-resolution cellular and tissue imaging, but optimizing resolution, speed, sample preservation, and signal-to-noise ratio simultaneously is challenging. The use of deep learning-based supersampling, known as point-scanning super-resolution (PSSR) imaging, can mitigate these limitations. PSSR facilitates high-resolution, fast, and sensitive image acquisition with otherwise unattainable resolution.
Article
Computer Science, Information Systems
Xunxiang Yao, Qiang Wu, Peng Zhang, Fangxun Bao
Summary: This study focuses on the importance of maintaining image roughness during image super-resolution, proposing an adaptive fractal interpolation function to address the issue. Experimental results demonstrate the effectiveness of this method in achieving high-performance super-resolution with minimal block artifacts.
IEEE TRANSACTIONS ON MULTIMEDIA
(2021)
Article
Geography, Physical
Tongwen Li, Huanfeng Shen, Qiangqiang Yuan, Liangpei Zhang
ISPRS JOURNAL OF PHOTOGRAMMETRY AND REMOTE SENSING
(2020)
Article
Meteorology & Atmospheric Sciences
Meiling Gao, Fei Chen, Huanfeng Shen, Huifang Li
THEORETICAL AND APPLIED CLIMATOLOGY
(2020)
Article
Environmental Sciences
Xuechen Zhang, Huanfeng Shen, Tongwen Li, Liangpei Zhang
INTERNATIONAL JOURNAL OF ENVIRONMENTAL RESEARCH AND PUBLIC HEALTH
(2020)
Article
Environmental Sciences
Chi Zhang, Huifang Li, Huanfeng Shen
Summary: A cirrus cloud correction method based on a scattering law was proposed to restore ground information in Landsat 8 OLI VNIR images by utilizing visible and cirrus bands. The method was evaluated on both simulated and acquired images, demonstrating its effectiveness in removing cirrus clouds and restoring ground information with high accuracy. Comparisons with existing methods showed superior results over various land and ocean scenes. Additionally, the method's applicability to other sensors with similar bands, such as the Sentinel-2 Multispectral Instrument, was confirmed.
REMOTE SENSING OF ENVIRONMENT
(2021)
Article
Geochemistry & Geophysics
Xiangchao Meng, Yiming Xiong, Feng Shao, Huanfeng Shen, Weiwei Sun, Gang Yang, Qiangqiang Yuan, Randi Fu, Hongyan Zhang
Summary: Pansharpening is a fundamental and active research topic in remote sensing that aims to sharpen low-spatial-resolution multispectral images using high-spatial-resolution panchromatic images. While performance evaluation is currently limited to individual images, data-driven approaches are gaining attention. The lack of publicly available benchmark datasets, especially large-scale ones, is a serious limitation for the future development of pansharpening methods.
IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE
(2021)
Article
Geochemistry & Geophysics
Xiangchao Meng, Gang Yang, Feng Shao, Weiwei Sun, Huanfeng Shen, Shutao Li
Summary: Despite the existence of many pansharpening methods, few are widely used in practice due to issues such as instability, implementation complexity, and time-consuming processes.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2022)
Article
Geochemistry & Geophysics
Hongyan Zhang, Jingyi Cai, Wei He, Huanfeng Shen, Liangpei Zhang
Summary: This article proposes a new method that simultaneously explores the low-rank characteristic of noise-free HSI and the low-rank structure of stripe noise on each band of the HSI, to achieve separation of noise-free HSI, stripe noise, and other mixed noise within one unified framework.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Computer Science, Artificial Intelligence
Jiang He, Jie Li, Qiangqiang Yuan, Huanfeng Shen, Liangpei Zhang
Summary: This article proposes an optimization-driven convolutional neural network (CNN) method for spectral super-resolution (SSR), utilizing auxiliary spectral response function (SRF) and channel attention module (CAM) to enhance spectral images reconstruction and classification.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2022)
Review
Geochemistry & Geophysics
Penghai Wu, Zhixiang Yin, Chao Zeng, Si-Bo Duan, Frank-Michael Gottsche, Xiaoshuang Ma, Xinghua Li, Hui Yang, Huanfeng Shen
Summary: High resolution remotely sensed land surface temperature is crucial for studying thermal environment, but spatial discontinuities can be introduced due to adverse atmospheric conditions, sensor malfunctioning, and scanning gaps.
IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE
(2021)
Article
Engineering, Electrical & Electronic
Shuang Luo, Huifang Li, Ruzhao Zhu, Yuting Gong, Huanfeng Shen
Summary: A novel network structure ESPFNet is proposed for salient shadow detection in aerial remote sensing images, achieved through a multitask learning framework. Experimental results demonstrate the effectiveness of ESPFNet method compared to existing methods in both qualitative and quantitative performance.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2021)
Article
Engineering, Electrical & Electronic
Peng Dou, Huanfeng Shen, Zhiwei Li, Xiaobin Guan, Wenli Huang
Summary: The study highlights the effectiveness of using a multiple classifier system to enhance remote sensing image classification accuracy. A novel deep-shallow learning framework is proposed, combining multiple classifier mechanism and deep learning architecture for improved classification accuracy.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2021)
Article
Engineering, Electrical & Electronic
Yao Shen, Huanfeng Shen, Qing Cheng, Liangpei Zhang
Summary: Satellite images are commonly used to monitor urban heat islands, with a focus on summer heat islands. Current approaches based on high and low spatial resolution images each have their limitations. A proposed solution integrates the advantages of both approaches to generate a comparable and detailed LST time series for analysis of the thermal environment.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2021)
Article
Geochemistry & Geophysics
Huanfeng Shen, Chenxia Zhou, Jie Li, Qiangqiang Yuan
Summary: The article introduces the recursive deep convolutional neural network (CNN) prior model for SAR image despeckling, which achieves better results in both quantitative and qualitative evaluations through end-to-end iterative training.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2021)
Proceedings Paper
Computer Science, Artificial Intelligence
Shuheng Zhao, Qiangqiang Yuan, Jie Li, Huanfeng Shen, Liangpei Zhang
IGARSS 2020 - 2020 IEEE INTERNATIONAL GEOSCIENCE AND REMOTE SENSING SYMPOSIUM
(2020)
Proceedings Paper
Computer Science, Artificial Intelligence
Qiang Zhang, Fujun Sun, Qiangqiang Yuan, Jie Li, Huanfeng Shen, Liangpei Zhang
IGARSS 2020 - 2020 IEEE INTERNATIONAL GEOSCIENCE AND REMOTE SENSING SYMPOSIUM
(2020)