IF 3.4 2区工程技术

Displays Pub Date : 2025-07-31 DOI: 10.1016/j.displa.2025.103174

Ruoyu Yang , Huaixin Chen , Sijie Luo , Zhixi Wang

{"title":"DELIA-Net: A Detail-Enhanced and Local Illumination Adjustment Network for nighttime road image enhancement","authors":"Ruoyu Yang , Huaixin Chen , Sijie Luo , Zhixi Wang","doi":"10.1016/j.displa.2025.103174","DOIUrl":"10.1016/j.displa.2025.103174","url":null,"abstract":"<div><div>Nighttime road image enhancement is vital for ensuring reliable visual input in low-light conditions, particularly in autonomous driving and intelligent traffic monitoring systems. However, existing methods often suffer from challenges such as detail loss, overexposed highlights, and color inconsistency, which significantly impair the performance of downstream perception and recognition tasks. To address these challenges, we propose a Detail-Enhanced and Local Illumination Adjustment Network (DELIA-Net). Specifically, we design a Detail-Enhanced U-Net (DEU-Net) that reinforces shallow-layer texture information and optimizes deep feature transmission, thereby improving the restoration of fine image structures. Second, we introduce a Bidirectional Difference Convolution (BDConv) module that extracts multi-directional edge features via horizontal and vertical convolutions, further augmented with an attention mechanism to enhance structural awareness. In addition, a Local Illumination Adjustment (LIA) attention module is proposed to adaptively enhance brightness in underexposed regions while suppressing overexposed areas, achieving balanced and natural illumination across the image. Finally, we incorporate a self-adaptive color loss that constrains inter-channel consistency and contrast within a single image, effectively mitigating unnatural color shifts during enhancement. To facilitate benchmarking in this field, we construct a real-world dataset comprising both nighttime and daytime road scenes, named NRRD (Night and Regular Road Dataset). Extensive experiments conducted on five datasets — NRRD, BDD100K, SICE, ExDark, and LIME — demonstrate that our method outperforms ten state-of-the-art approaches in terms of image quality, detail preservation, and color restoration, validating its effectiveness and generalization capability under various low-light conditions.</div></div>","PeriodicalId":50570,"journal":{"name":"Displays","volume":"90 ","pages":"Article 103174"},"PeriodicalIF":3.4,"publicationDate":"2025-07-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144749842","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

CurvDrop: Data-efficient learning for medical image segmentation 曲率drop：用于医学图像分割的数据高效学习

IF 3.4 2区工程技术

Displays Pub Date : 2025-07-30 DOI: 10.1016/j.displa.2025.103170

Xian-Tao Wu , Xiao-Diao Chen , Hong-Yu Chen , Wen Wu

{"title":"CurvDrop: Data-efficient learning for medical image segmentation","authors":"Xian-Tao Wu , Xiao-Diao Chen , Hong-Yu Chen , Wen Wu","doi":"10.1016/j.displa.2025.103170","DOIUrl":"10.1016/j.displa.2025.103170","url":null,"abstract":"<div><div>More training data can frequently boost the performance, but also be accompanied by increased training costs and reliance on high-performance computing resources. This work concerns lots of similar and less-informative slices within the medical volumes, and aims to drop out some of them (referred as redundancy) while maintaining the model’s performance. We observe that the curvatures of loss surface are highly relevant to the data redundancy: Samples with low curvatures tend to be organ-sparse and cannot produce enough loss to guide the optimization process even at early training phrase; while samples with high curvature usually have ambiguous and unrepresentative contents, which is essential to boost the generalization ability. From these observations, we introduce CurvDrop, a sample selection and training algorithm for data efficient training. Specifically, we first calculate the curvatures for all training samples with Hessian from a warmed-up model. Then, a condensed dataset can be constructed by dropping out partial samples with lowest curvatures. To avoid overfitting and maintain the robustness, we design a curvature-assisted regularization to facilitate the flatness of loss surface on a new segmentation network. A unique advantage of our method is that the condensed set identified by a specific network can be generalized across other architectures. Experiments on a widely used medical segmentation benchmark (<em>i.e.</em>, Synapse and ACDC dataset) show that our CurvDrop can save 24% of training costs by using only 70% training data to achieve 98.8% performance. Code and model will be available at <span><span>https://github.com/wuwen1994/CurvDrop</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":50570,"journal":{"name":"Displays","volume":"90 ","pages":"Article 103170"},"PeriodicalIF":3.4,"publicationDate":"2025-07-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144739497","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Integrating bio-signal recording and playback for mixed reality asynchronous collaboration 为混合现实异步协作集成生物信号记录和回放

IF 3.4 2区工程技术

Displays Pub Date : 2025-07-30 DOI: 10.1016/j.displa.2025.103171

Bokai Zheng , Zhuang Chang , Zirui Xiao , Weiping He , Mark Billinghurst

引用次数: 0

DRIGNet: Low-light image enhancement based on dual-range information guidance DRIGNet：基于双距离信息制导的微光图像增强

IF 3.7 2区工程技术

Displays Pub Date : 2025-07-26 DOI: 10.1016/j.displa.2025.103163

Feng Huang, Jiong Huang, Jing Wu, Jianhua Lin, Jing Guo, Yunxiang Li, Zhewei Liu

{"title":"DRIGNet: Low-light image enhancement based on dual-range information guidance","authors":"Feng Huang, Jiong Huang, Jing Wu, Jianhua Lin, Jing Guo, Yunxiang Li, Zhewei Liu","doi":"10.1016/j.displa.2025.103163","DOIUrl":"10.1016/j.displa.2025.103163","url":null,"abstract":"<div><div>The task of low-light image enhancement aims to reconstruct details and visual information from degraded low-light images. However, existing deep learning methods for feature processing usually lack feature differentiation or fail to implement reasonable differentiation handling, which can limit the quality of the enhanced images, leading to issues like color distortion and blurred details. To address these limitations, we propose Dual-Range Information Guidance Network (DRIGNet). Specifically, we develop an efficient U-shaped architecture Dual-Range Information Guided Framework (DGF). DGF decouples traditional image features into dual-range information while integrating stage-specific feature properties with the proposed dual-range information. We design the Global Dynamic Enhancement Module (GDEM) using channel interaction and the Detail Focus Module (DFM) with three-directional filter, both embedded in DGF to model long-range and short-range features respectively. Additionally, we introduce a feature fusion strategy, Attention-Guided Fusion Module (AGFM), which merges dual-range information, facilitating complementary enhancement. In the encoder, DRIGNet extracts coherent long-range information and enhances the global structure of the image; in the decoder, DRIGNet captures short-range information and fuse dual-rage information to restore detailed areas. Finally, we conduct extensive quantitative and qualitative experiments to demonstrate that the proposed DRIGNet outperforms the current State-of-the-Art (SOTA) methods across ten datasets.</div></div>","PeriodicalId":50570,"journal":{"name":"Displays","volume":"90 ","pages":"Article 103163"},"PeriodicalIF":3.7,"publicationDate":"2025-07-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144714096","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A halftone image quality assessment method based on gradient and texture 基于梯度和纹理的半色调图像质量评价方法

IF 3.4 2区工程技术

Displays Pub Date : 2025-07-26 DOI: 10.1016/j.displa.2025.103165

Xinhong Zhang , Jiayin Zhao , Fan Zhang

引用次数: 0

Omnidirectional image quality assessment with gated dual-projection fusion 基于门控双投影融合的全向图像质量评估

IF 3.7 2区工程技术

Displays Pub Date : 2025-07-24 DOI: 10.1016/j.displa.2025.103173

ChengZhi Xiao, RuiKang Yu

{"title":"Omnidirectional image quality assessment with gated dual-projection fusion","authors":"ChengZhi Xiao, RuiKang Yu","doi":"10.1016/j.displa.2025.103173","DOIUrl":"10.1016/j.displa.2025.103173","url":null,"abstract":"<div><div>Existing omnidirectional image quality assessment (OIQA) models typically rely on the equirectangular projection (ERP) or cubemap projection (CMP) of omnidirectional images as inputs. However, the deformation in ERP and the discontinuities at the boundaries of CMP limit the network’s ability to represent image information, leading to information loss. Therefore, it is necessary to fuse these two projections of omnidirectional images to achieve comprehensive feature representation. Current OIQA models only integrate and interact high-level features extracted from different projection formats at the last stage of the network, overlooking potential information loss at each stage within the network. To this end, we consider the respective strengths and weaknesses of the two projections, and design a feature extraction and fusion module at each stage of the network to enhance the model’s representation capability. Specifically, the ERP features are first decomposed into two projection formats before being fed into each feature extraction stage of the network for separate processing. Subsequently, we introduce the gating mechanism and develop a Gated Dual-Projection Fusion module (GDPF) to interactively fuse the features computed from both the ERP and CMP projection formats. GDPF allows the developed model to enhance critical information while filtering out deformation and discontinuous information. The fused features are then input into the next stage, where the aforementioned operations are repeated. This process alleviates the issues of feature representation caused by deformation in ERP and discontinuities in CMP and the fused features are used for quality prediction. Experiments on three public datasets demonstrate the superior prediction accuracy of the proposed model.</div></div>","PeriodicalId":50570,"journal":{"name":"Displays","volume":"90 ","pages":"Article 103173"},"PeriodicalIF":3.7,"publicationDate":"2025-07-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144704424","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A joint learning framework for fake news detection 假新闻检测联合学习框架

IF 3.7 2区工程技术

Displays Pub Date : 2025-07-23 DOI: 10.1016/j.displa.2025.103154

Muhammad Abdullah , Zan Hongying , Arifa Javed , Orken Mamyrbayev , Fabio Caraffini , Hassan Eshkiki

引用次数: 0

LR-Inst: A lightweight and robust instance segmentation network for apple detection in complex orchard environments 一个轻量级和鲁棒的实例分割网络，用于复杂果园环境中的苹果检测

IF 3.7 2区工程技术

Displays Pub Date : 2025-07-21 DOI: 10.1016/j.displa.2025.103156

Hengrong Guo , Hao Wan , Xilei Zeng, Han Zhang, Zeming Fan

{"title":"LR-Inst: A lightweight and robust instance segmentation network for apple detection in complex orchard environments","authors":"Hengrong Guo , Hao Wan , Xilei Zeng, Han Zhang, Zeming Fan","doi":"10.1016/j.displa.2025.103156","DOIUrl":"10.1016/j.displa.2025.103156","url":null,"abstract":"<div><div>Apple instance segmentation is a critical task in the implementation of automated harvesting systems. Despite significant advances in instance segmentation, current methods remain impractical for deployment due to their architectural complexity and slow inference speeds. While lightweight models have been introduced to improve efficiency, their performance degrades in orchard environments under occlusion, fruit overlap, and varying lighting conditions. To address these challenges, we present LR-Inst, a lightweight and robust instance segmentation network. First, we design an innovative cross-level feature fusion architecture that exploits the rich spatial details and semantic information present in intermediate-layer features. Then, a set of efficient modules is designed to further boost feature representation, including the Spatial-Semantic Feature Fusion Module (SSFM), the Dynamic Spatial-Semantic Fusion Module (DSSFM), the Feature Aggregation and Shuffle Module (FASM), and the Channel-Spatial Attention Module (CSAM). Experimental results demonstrate that LR-Inst contains only 3.742M parameters and requires 8.581G FLOPs. When evaluated on our self-collected orchard dataset, LR-Inst achieves a detection average precision (AP) of 0.946 and a segmentation AP of 0.944, outperforming several state-of-the-art (SOTA) models.</div></div>","PeriodicalId":50570,"journal":{"name":"Displays","volume":"90 ","pages":"Article 103156"},"PeriodicalIF":3.7,"publicationDate":"2025-07-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144696677","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

The application of VR in interior design education to enhance design effectiveness and student experience VR在室内设计教育中的应用，提高设计效果和学生体验

IF 3.7 2区工程技术

Displays Pub Date : 2025-07-21 DOI: 10.1016/j.displa.2025.103161

Pengjun Wu , Wencui Zhang , Peiyuan Li , Yao Liu

{"title":"The application of VR in interior design education to enhance design effectiveness and student experience","authors":"Pengjun Wu , Wencui Zhang , Peiyuan Li , Yao Liu","doi":"10.1016/j.displa.2025.103161","DOIUrl":"10.1016/j.displa.2025.103161","url":null,"abstract":"<div><div>As Interior Design Education (IDE) evolves to meet increasingly complex and diverse demands, traditional teaching methods face limitations in areas such as design presentation, teacher–student interaction, and spatial perception, often leading to reduced learning effectiveness. Virtual Reality (VR), with its immersive and interactive features, offers promising solutions to these challenges. This study developed a VR-based interior design education platform incorporating Level of Detail (LOD) technology to improve instructional precision and learning outcomes. To systematically evaluate the teaching effectiveness, the study employed evaluation indicators grounded in the Technology Acceptance Model (TAM), emphasizing perceived usefulness and perceived ease of use as key dimensions influencing learners’ acceptance of VR technology. Specifically, content comprehensiveness, visual clarity, and spatial understanding were selected as core evaluation metrics reflecting these TAM constructs. An experimental comparison with traditional teaching methods assessed these dimensions. Results showed the VR-based approach significantly outperformed traditional methods, with higher average scores in comprehensiveness (90.68 ± 4.00 vs. 82.35 ± 2.20), visibility (91.08 ± 4.11 vs. 83.66 ± 3.85), and spatial effects (92.98 ± 3.22 vs. 85.64 ± 3.96). These findings highlight the advantages of LOD-enhanced VR teaching in improving clarity and interaction efficiency. Focus group interviews further confirmed its effectiveness in enhancing students’ understanding and communication.</div></div>","PeriodicalId":50570,"journal":{"name":"Displays","volume":"90 ","pages":"Article 103161"},"PeriodicalIF":3.7,"publicationDate":"2025-07-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144685861","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Orthogonal translation computed laminography reconstruction based on self-prior information and adaptive weighted total variation 基于自先验信息和自适应加权总变差的正交平移计算层析重建

IF 3.7 2区工程技术

Displays Pub Date : 2025-07-20 DOI: 10.1016/j.displa.2025.103169

Chuandong Tan , Chao Long , Yarui Xi , Zhiting Chen , Xinxin Lin , Fenglin Liu , Yufang Cai , Liming Duan

{"title":"Orthogonal translation computed laminography reconstruction based on self-prior information and adaptive weighted total variation","authors":"Chuandong Tan , Chao Long , Yarui Xi , Zhiting Chen , Xinxin Lin , Fenglin Liu , Yufang Cai , Liming Duan","doi":"10.1016/j.displa.2025.103169","DOIUrl":"10.1016/j.displa.2025.103169","url":null,"abstract":"<div><div>Orthogonal translation computed laminography (OTCL) provides an effective non-destructive testing method for plate-like objects. Nevertheless, OTCL images suffer from aliasing artifacts due to the inherent incompleteness of projection data, negatively impacting flaw characterization, dimensional metrology, and failure analysis. To reveal the cause of aliasing artifacts, the three-dimensional frequency domain characteristics of OTCL are analyzed. We further propose a novel reconstruction algorithm to mitigate aliasing artifacts, termed self-prior information guidance and adaptive weight total variation constraint (SPIG-AwTV). The SPIG-AwTV comprises two components: a self-prior information guidance (SPIG) regularization term and an adaptive weighted total variation (AwTV) regularization term. Specifically, SPIG is derived from filtered backprojection reconstruction result via contour extraction and masking. The AwTV regularization term is tailored to the gradient features of OTCL images in different directions. Experimental results demonstrate that the SPIG-AwTV outperforms existing methods in suppressing aliasing artifacts, preserving edges, and achieving higher-quality OTCL images.</div></div>","PeriodicalId":50570,"journal":{"name":"Displays","volume":"90 ","pages":"Article 103169"},"PeriodicalIF":3.7,"publicationDate":"2025-07-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144685992","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Displays最新文献