Article
Neurosciences
Tianshu Song, Leida Li, Hancheng Zhu, Jiansheng Qian
Summary: Current image quality assessment metrics mainly focus on distortion aspects, neglecting the importance of intelligibility for robust quality estimation. This study proposes a new framework for integrating intelligibility to build a highly generalizable image quality model, achieving better performance than state-of-the-art metrics. Feature selection strategy is devised to avoid negative transfer during the fusion process.
FRONTIERS IN NEUROSCIENCE
(2021)
Article
Computer Science, Information Systems
Jiachen Yang, Yanshuang Zhou, Yang Zhao, Jiabao Wen
Summary: This paper proposes a quality assessment method for tone-mapped images based on generating multi-exposure sequences. By using a generative adversarial network to generate sequences with different exposure levels and utilizing a convolutional neural network to extract features and learn mapping relationships, the proposed method achieves quality assessment of tone-mapped images.
JOURNAL OF VISUAL COMMUNICATION AND IMAGE REPRESENTATION
(2022)
Article
Chemistry, Multidisciplinary
Jihyoung Ryu
Summary: The purpose of NR-IQA is to measure perceived image quality based on subjective judgments, but this is a complicated challenge due to the lack of a clean reference image. A unique hybrid model is presented in this research, which leverages both pre-trained CNN and the unified learning mechanism to handle the NR-IQA challenge by extracting local and non-local characteristics. Deep analysis of the proposed framework demonstrates improved monotonicity relationship between objective and subjective ratings. The suggested technique outperforms current state-of-the-art NR-IQA measures, as shown in analyses of the largest NR-IQA benchmark datasets.
APPLIED SCIENCES-BASEL
(2023)
Article
Mathematics
Pei-Fen Tsai, Huai-Nan Peng, Chia-Hung Liao, Shyan-Ming Yuan
Summary: To improve data transmission efficiency, image compression is commonly used, but it causes image distortion. The generative adversarial network (GAN)-based method is an advanced algorithm with a high correlation to the human visual system (HVS). We proposed an ensemble image quality assessment (IQA) called ATDIQA to evaluate the performance of GAN-based IR algorithms, which combines multiscale features and local features to give weights on transformers and convolutional neural network (CNN) IQA. ATDIQA not only performs well on the PIPAL dataset of GAN IR algorithms but also has good model generalization over traditional distorted image datasets like LIVE and TID2013. It demonstrates a high correlation with human judgment score of distorted images.
Article
Engineering, Electrical & Electronic
Yuchen Zou, Chengcheng Liu, Huikai Shao, Dexing Zhong
Summary: This article proposes a label-free palmprint image quality assessment framework using pseudo-label generation and ranking guidance. It aims to determine the trustworthiness of palmprint images input to the system and improve the performance of palmprint recognition models.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2023)
Article
Chemistry, Multidisciplinary
Jihyoung Ryu
Summary: Deep learning has been extensively studied in blind image quality assessment (BIQA), but the scarcity of high-quality algorithms hinders further development and real-time application. This study proposes a patch-based technique with a visual saliency module to predict image quality by learning only important information. Using the Inception-ResNet-v2 neural network architecture and benchmark database evaluation, the proposed strategy demonstrates better performance than popular IQA approaches.
APPLIED SCIENCES-BASEL
(2022)
Article
Computer Science, Information Systems
Nandhini Chockalingam, Brindha Murugan
Summary: The article introduces the methods of dense convolution network (DSC-Net) and multimodal dense convolution network (MDSC-Net) for image quality assessment. DSC-Net improves the quality of image representation by reducing the number of parameters and addressing the issue of overfitting. MDSC-Net combines texture features and spatial features to enhance the performance of image quality prediction through multimodal data.
FRONTIERS OF INFORMATION TECHNOLOGY & ELECTRONIC ENGINEERING
(2023)
Article
Computer Science, Information Systems
Yang Yang, Yingqiu Ding, Ming Cheng, Weiming Zhang
Summary: This article proposes a no-reference quality assessment algorithm for contrast-distorted images based on gray and color-gray-difference (CGD) space. The method considers both local and global features in the gray space, using the distribution characteristics of the grayscale histogram to represent global features and the fusion of Local Binary Pattern (LBP) operator and gradient to describe local features. In the CGD space, quality perception features are extracted from randomly extracted patches and used to train the prediction model using AdaBoosting back propagation (BP) neural network. Experimental results on five contrast-related image databases have demonstrated the superiority of this method compared with recent related algorithms.
ACM TRANSACTIONS ON MULTIMEDIA COMPUTING COMMUNICATIONS AND APPLICATIONS
(2023)
Article
Computer Science, Artificial Intelligence
Yuan Chen, Yang Zhao, Li Cao, Wei Jia, Xiaoping Liu
Summary: This paper proposes a blind cartoon image quality assessment method based on convolutional neural networks, and improves the network's robustness by establishing a large-scale dataset and implementing a random degradation strategy. Experimental results demonstrate the effectiveness and robustness of the proposed method on synthetic and real-world cartoon image datasets.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Computer Science, Information Systems
Zhenjun Tang, Zhiyuan Chen, Zhixin Li, Bineng Zhong, Xianquan Zhang
Summary: Image Quality Assessment (IQA) is a critical task of computer vision. In this paper, a novel method called UniDASTN is proposed for FR-IQA, which combines the Dual-Attention and Siamese Transformer Network. The proposed method effectively evaluates the distortion in distorted images through the spatial attention module and the dual-attention strategy. The experiments on standard IQA databases demonstrate that UniDASTN outperforms some state-of-the-art FR-IQA methods.
ACM TRANSACTIONS ON MULTIMEDIA COMPUTING COMMUNICATIONS AND APPLICATIONS
(2023)
Article
Computer Science, Artificial Intelligence
Yannan Zheng, Weiling Chen, Rongfu Lin, Tiesong Zhao, Patrick Le Callet
Summary: Due to the complex lighting environment underwater, image quality is often compromised by scattering, warping, and noise. To address this, underwater image enhancement techniques have been studied, and a new objective evaluation metric called Underwater Image Fidelity (UIF) is proposed. The UIF metric utilizes statistical features in CIELab space to measure naturalness, sharpness, and structure indexes, which together provide a comprehensive assessment of the enhanced underwater image quality.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2022)
Article
Computer Science, Information Systems
Junyong You, Jari Korhonen
Summary: This paper proposes an AIHIQnet architecture for no-reference quality assessment, which combines hierarchical image quality perception, attention, and contrast sensitivity mechanisms to accurately evaluate the quality of natural images.
JOURNAL OF VISUAL COMMUNICATION AND IMAGE REPRESENTATION
(2022)
Article
Engineering, Electrical & Electronic
Yue Liu, Zhangkai Ni, Shiqi Wang, Hanli Wang, Sam Kwong
Summary: In this paper, a novel and effective image quality assessment (IQA) algorithm called local-global frequency feature-based model (LGFM) is proposed for high dynamic range (HDR) images. The algorithm extracts local and global frequency features using Gabor and Butterworth filters applied to the luminance component of the HDR image. Similarity measurement and feature pooling strategy are performed on the frequency features to obtain a predicted single quality score. Experimental results on four benchmarks demonstrate that LGFM provides higher consistency with subjective perception compared to state-of-the-art HDR IQA methods.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2023)
Article
Geochemistry & Geophysics
Mengyao Li, Yufei Lin, Liquan Shen, Zheyin Wang, Kun Wang, Zhengyong Wang
Summary: This research proposes a human perceptual quality-driven underwater image enhancement framework, which trains the model on real datasets and takes human visual perception into account. By introducing a novel attention mechanism and confidence map, the proposed framework achieves superior results in underwater image enhancement compared to other methods.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Computer Science, Information Systems
Pengfei Guo, Lang He, Shuangyin Liu, Delu Zeng, Hantao Liu
Summary: This paper investigates the performance of five popular enhancement algorithms for underwater images and analyzes their impact on perceptual quality. It also evaluates the visual quality objectively, aiming to develop objective metrics for automatic assessment of underwater image enhancement quality.
IEEE TRANSACTIONS ON MULTIMEDIA
(2022)
Article
Engineering, Electrical & Electronic
Xinpeng Huang, Yilei Chen, Ping An, Liquan Shen
Summary: In this paper, a prediction-oriented disparity rectification (PoDR) model is proposed to address the limitation of geometry-based light field compression methods. The PoDR model utilizes the 4D structural prior of light fields and practical non-uniform disparity distribution to improve prediction accuracy and compression performance. Experimental results show that the proposed PoDR model achieves better light field fundamental capability compared to state-of-the-art methods.
IEEE TRANSACTIONS ON BROADCASTING
(2023)
Article
Engineering, Electrical & Electronic
Mengyao Li, Kun Wang, Liquan Shen, Yufei Lin, Zhengyong Wang, Qijie Zhao
Summary: Artificial light is often used to assist underwater photography due to degradation of underwater images. However, it alters the normal imaging process and results in non-uniform illumination and color distortion. Existing enhancement algorithms fail to address these issues. Therefore, a novel algorithm based on luminance correction and artificial light area color restoration is proposed.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2023)
Article
Engineering, Electrical & Electronic
Mengyao Li, Liquan Shen, Yufei Lin, Kun Wang, Jinbo Chen
Summary: Due to the limited bandwidth of underwater wireless acoustic channel, underwater images (UWIs) require higher compression ratio than terrestrial images. Existing image compression methods cannot fulfill practical applications that require detailed observation of foreground objects (FGOs) in UWIs while only general viewing of the background. To overcome this limitation, we propose an underwater physical prior-based extreme compression network (PPECN) for UWIs compression.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2023)
Article
Engineering, Electrical & Electronic
Zheyin Wang, Liquan Shen, Zhengyong Wang, Yufei Lin, Yanliang Jin
Summary: In this paper, a novel mapping-based method is proposed to evaluate the quality of underwater enhanced images. The proposed GLCQE model generates reference images and utilizes multiple networks to assess distortions and quality. Experimental results demonstrate the superiority of this model compared to other approaches.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2023)
Article
Engineering, Electrical & Electronic
Zhengkai Fang, Liquan Shen, Mengyao Li, Zhengyong Wang, Yanliang Jin
Summary: Machine analysis of underwater images is crucial for most underwater applications, but communication bandwidth limitations and underwater degradation make accurate machine recognition challenging. To address this, we propose a novel underwater image compression framework that utilizes underwater priors to enhance degraded features through contrastive learning and efficiently compress machine-friendly features under low bit-rates.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2023)
Article
Engineering, Civil
Zhengkai Fang, Liquan Shen, Mengyao Li, Zhengyong Wang, Yanliang Jin
Summary: A low bit-rate compression is needed for underwater images due to limited bandwidth in underwater acoustic communication. However, existing compression methods fail to consider unique characteristics of underwater images, such as color shift and haze effect. To address this, we propose an extreme underwater image compression framework that utilizes underwater priors to provide scalability for machine vision and human vision. Experimental results show the superiority of our framework in machine vision tasks and perception quality compared to traditional compression methods and learned-based methods.
IEEE JOURNAL OF OCEANIC ENGINEERING
(2023)
Article
Engineering, Electrical & Electronic
Lihao Zhuang, Liquan Shen, Zhengyong Wang, Yinyi Li
Summary: This paper proposes a novel priors guided adaptive underwater compressive sensing framework, dubbed UCSNet, which can effectively sample and reconstruct underwater images under a fixed low sampling ratio. The framework consists of three sub-networks: underwater priors extraction and guidance network, sampling matrix generation network, and channel-wise reconstruction network. Experimental results demonstrate that our framework outperforms other state-of-the-art methods in terms of underwater image reconstruction quality.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2023)
Article
Computer Science, Hardware & Architecture
Zichi Wang, Guorui Feng, Liquan Shen, Xinpeng Zhang
Summary: Existing cover selection methods in steganography focus on embedding distortion but overlook image similarity. This paper proposes a new cover selection method that combines image similarity based on SVD and embedding distortion calculation. The obtained image similarity and embedding distortion are combined to form a new cover selection strategy, outperforming state-of-the-art methods in steganalysis tests.
IEEE TRANSACTIONS ON DEPENDABLE AND SECURE COMPUTING
(2023)
Article
Geochemistry & Geophysics
Hui Long, Liquan Shen, Zhengyong Wang, Jinbo Chen
Summary: In this study, a novel underwater forward-looking sonar image detection network (UFIDNet) is proposed to address the challenges of multiplicative speckle noise and scene prior in FLS images. The network includes a speckle reduction auxiliary branch and a feature selection strategy to improve detection performance. Experimental results demonstrate that UFIDNet outperforms state-of-the-art detectors on real FLS datasets.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Computer Science, Artificial Intelligence
Zhengyong Wang, Liquan Shen, Mai Xu, Mei Yu, Kun Wang, Yufei Lin
Summary: Recently, learning-based algorithms have made great progress in underwater image enhancement. However, existing methods fail to consider the domain gap between synthetic and real data, leading to poor generalization in real-world underwater scenarios. Additionally, the complex underwater environment results in a distribution gap among real data itself. In this study, we propose a novel Two-phase Underwater Domain Adaptation network (TUDA) to bridge both the inter-domain and intra-domain gaps. Extensive experiments show that TUDA outperforms existing works in terms of visual quality and quantitative metrics.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2023)
Article
Computer Science, Information Systems
Xiangyu Hu, Liquan Shen, Mingxing Jiang, Ran Ma, Ping An
Summary: This study proposes a light adaptation HDR recovery framework (LA-HDR), which achieves robust HDR recovery under different light conditions by generating multiple images and fusing details. The proposed framework eliminates high-light recovery artifacts and shows the best average performance among tested state-of-art HDR recovery methods, with minimal influence from input light conditions.
IEEE TRANSACTIONS ON MULTIMEDIA
(2023)
Article
Geochemistry & Geophysics
Mengyao Li, Yufei Lin, Liquan Shen, Zheyin Wang, Kun Wang, Zhengyong Wang
Summary: This research proposes a human perceptual quality-driven underwater image enhancement framework, which trains the model on real datasets and takes human visual perception into account. By introducing a novel attention mechanism and confidence map, the proposed framework achieves superior results in underwater image enhancement compared to other methods.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Engineering, Electrical & Electronic
Chao Yang, Ping An, Liquan Shen
Summary: This article introduces a data-driven transform-based feature enhancement method for blind image quality measurement, combining the advantages of hand-crafted features and learning-based features. The proposed method achieves highly competitive performance on various databases.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2022)
Article
Engineering, Electrical & Electronic
Mingxing Jiang, Liquan Shen, Min Hu, Ping An, Yu Gu, Fuji Ren
Summary: Measuring electronic display quality is crucial in current consumer displays. This study focuses on establishing new IQM protocols to differentiate the display quality of SDR electronic devices, particularly in tone mapping. An IQM model is proposed to exhibit perceptual attributes and artifacts unique to tone mapping, and various techniques are used to characterize the overall image quality. The results show that using IQM protocols helps accurately measure the quality during tone mapping, facilitating optimal display.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2022)