Article
Chemistry, Analytical
Ivana Shopovska, Ana Stojkovic, Jan Aelterman, David Van Hamme, Wilfried Philips
Summary: Intelligent driver assistance systems are increasingly popular and have the ability to detect vulnerable road users. However, standard imaging sensors perform poorly in strong illumination contrast conditions. This study focuses on the use of HDR imaging sensors and the need for tone mapping in vehicle perception systems. The proposed DI-TM method achieves the best performance in terms of detection metrics in challenging dynamic range conditions, with a 13% improvement compared to existing methods.
Article
Computer Science, Information Systems
Imran Mehmood, Xinye Shi, Muhammad Usman Khan, Ming Ronnier Luo
Summary: One of the key challenges in tone mapping is to preserve the perceptual quality of high dynamic range (HDR) images when mapping them to standard dynamic range (SDR) displays. This research proposes a new tone mapping operator (TMOz) that leverages CIECAM16 perceptual attributes to achieve more optimal colorfulness reproduction. Both objective and subjective evaluations show that the proposed model outperforms existing TMO algorithms.
Article
Chemistry, Analytical
Yumei Li, Ningfang Liao, Wenmin Wu, Chenyang Deng, Yasheng Li, Qiumei Fan, Chuanjie Liu
Summary: This study aims to solve the difficulties in displaying high dynamic range (HDR) images on conventional standard display devices. The proposed modified tone-mapping operator (TMO), called iCAM06-m, combines iCAM06 and a multi-scale enhancement algorithm to compensate for saturation and hue drift. Subjective evaluation shows the better performance of iCAM06-m compared to other TMOs, confirming the effectiveness of chroma compensation and multi-scale decomposition. The proposed algorithm overcomes the limitations of other algorithms and is a good candidate for a general purpose TMO.
Review
Engineering, Electrical & Electronic
Xueyu Han, Ishtiaq Rasool Khan, Susanto Rahardja
Summary: Compared to conventional digital imaging, high dynamic range (HDR) image tone mapping technology faithfully reproduces real-world scenes on common displays for enhanced viewing experience. This paper presents a literature review and benchmarks of existing tone mapping methods, providing updated information and paving the way for further developments in the field.
DIGITAL SIGNAL PROCESSING
(2023)
Article
Computer Science, Information Systems
Li Li, Kaiyi Zhao, Jiangzhang Gan, Saihua Cai, Tong Liu, Huiyu Mu, Ruizhi Sun
Summary: The paper introduces a semi-supervised learning method SS-GSELM, which combines dynamic graph learning and self-paced learning mechanism. It selects samples with smaller loss values for learning and gradually adds more samples, while using local consistency property and constructing adaptive graph matrix to enhance the performance of the learning machine.
INFORMATION PROCESSING & MANAGEMENT
(2021)
Article
Engineering, Electrical & Electronic
Pedram Mohammadi, Mahsa T. Pourazad, Panos Nasiopoulos
Summary: The introduction of High Dynamic Range (HDR) technologies has brought about significant visual improvements and opened up new markets for a wide range of industries. Owners of legacy Standard Dynamic Range (SDR) content now have the opportunity to convert their content to the HDR standard and take advantage of the enhanced capabilities of HDR displays. A novel high visual quality video inverse Tone Mapping Operator (iTMO) has been proposed, which is based on human visual perception and employs a segmentation method according to the Human Visual System (HVS) sensitivity, resulting in high visual quality HDR videos.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2021)
Article
Mathematical & Computational Biology
Ishtiaq Rasool Khan, Susanto Rahardja
Summary: This paper proposes a unified structure that can represent any global tone-mapping algorithm with a small array of coefficients. The unified implementation can convert HDR images and videos to LDR in real time, producing identical LDR images to the original algorithm. It has low and fixed execution time, independent of algorithm and content type.
MATHEMATICAL BIOSCIENCES AND ENGINEERING
(2022)
Article
Computer Science, Information Systems
Nam Hoang Nguyen, Tu Van Vo, Chul Lee
Summary: This study introduces an optimized tone-mapping algorithm based on human visual system response model to preserve perceptual responses between HDR and SDR devices, improving image quality. By efficiently solving optimization problems to obtain the optimal tone-mapping curve, experimental results demonstrate the algorithm's superiority in subjective and objective evaluations compared to conventional algorithms.
Article
Automation & Control Systems
Yichen Fu, Junfeng Fan, Fengshui Jing, Min Tan
Summary: This article proposes a set of hardware devices and region adaptive structured light algorithms to overcome the problems caused by high surface reflectivity in fringe projection profilometry. The proposed method is efficient, superior, and accurate in measuring objects with high reflective surface properties.
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS
(2023)
Article
Optics
Yi-Zheng Lang, Yun-Sheng Qian, Xiang-Yu Kong, Jing-Zhi Zhang
Summary: An adaptive parameter-free and scene-adaptive TMO method is proposed to adjust the brightness and contrast of HDR images, decompose the images into detail and base layers, and retain as much scene information as possible, resulting in high-quality tone-mapped images.
Article
Engineering, Electrical & Electronic
Pei Zhou, Yue Cheng, Jiangping Zhu, Jialing Hu
Summary: This article proposes a novel HDR 3D surface measurement method based on adaptive speckle projection, which takes into account the reflection distribution. By establishing an efficient segmentation-based mapping strategy and optimizing projection intensity, the method achieves fast generation of adaptive speckle patterns and demonstrates clear advantages in measurement accuracy and coverage compared to traditional methods.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2023)
Article
Computer Science, Information Systems
Liwen Liu, Zhong Wang, Jianbin Bai, Xiangfeng Yang, Yunchuan Yang, Jianbo Zhou
Summary: This study proposes an ensemble self-paced learning approach with an adaptive mixture weighting mechanism, achieving the best results in experiments by adjusting parameters.
Article
Environmental Sciences
Jifa Chen, Guojun Zhai, Gang Chen, Bo Fang, Ping Zhou, Nan Yu
Summary: This paper proposes a category-level adaptive method for coastal land cover mapping, utilizing an adversarial framework to align semantic features across image domains, effectively addressing the complex spatial details and low inter-class variances of coastal objects.
Article
Computer Science, Information Systems
Yifei Huang, Sheng Qiu, Changbo Wang, Chenhui Li
Summary: This paper proposes an innovative high-dynamic-range image color transfer generative adversarial network (HDRCTGAN) that learns fine image representations through self-supervised learning for transferring colors from reference images to target images. The method requires only unlabeled HDR images for training, instead of supervised learning with many ground truth pairs, resulting in pleasing visual results.
IEEE TRANSACTIONS ON MULTIMEDIA
(2021)
Article
Computer Science, Information Systems
Yuyang Ren, Haonan Zhang, Peng Yu, Luoyi Fu, Xinde Cao, Xinbing Wang, Guihai Chen, Fei Long, Chenghu Zhou
Summary: Self-supervised graph-level representation learning, namely Ada-MIP, considers both the unique features and common features of graphs, by maximizing the agreement between graph representations from different views and calculating the proximity among graphs using graph kernels. Ada-MIP learns adaptive views through a learnable and probably injective augmenter, and integrates graphs' unique and common information into the learned graph representations. It achieves superior performance in both unsupervised and semi-supervised tasks.
ACM TRANSACTIONS ON KNOWLEDGE DISCOVERY FROM DATA
(2023)
Article
Computer Science, Information Systems
Shuhui Gong, John Cartlidge, Ruibin Bai, Yang Yue, Qingquan Li, Guoping Qiu
Summary: The Huff model is calibrated using GTWR to show significant geographical and temporal variations in attractiveness and travel cost parameters, with wealthy customers being more sensitive to a shopping centre's attractiveness. Factors such as customer wealth, spare time, and travel mode influence shopping behaviors, and there are differences in customer behaviors between New York and Shenzhen, particularly at weekends. The GTWR calibration and identification of factors affecting urban travel behaviors can contribute to optimizing urban transportation design.
Article
Computer Science, Theory & Methods
Hui Yin, Yuanhao Gong, Guoping Qiu
Summary: The article introduces a new local window based image processing framework called combined window filtering (CWF), which combines full window filtering strategy (FWF) and side window filtering strategy (SWF) to achieve improved edge-preserving and texture-removing capabilities. By using different filtering strategies for edges and textures, the new framework significantly enhances the performance in applications.
MULTIDIMENSIONAL SYSTEMS AND SIGNAL PROCESSING
(2021)
Article
Computer Science, Artificial Intelligence
Qing Li, Jiasong Zhu, Rui Cao, Ke Sun, Jonathan M. Garibaldi, Qingquan Li, Bozhi Liu, Guoping Qiu
Summary: This study proposes a relative geometry-aware Siamese neural network, which enhances the performance of deep learning methods by explicitly leveraging the relative geometry constraints between images. Multi-task learning is used to predict absolute and relative poses, while shared-weight twin networks are regularized to ensure correct estimations globally and locally. Additionally, an adaptive metric distance loss is designed to learn features capable of distinguishing poses of visually similar images from different locations.
Article
Computer Science, Interdisciplinary Applications
Ruitao Xie, Jingxin Liu, Rui Cao, Connor S. Qiu, Jiang Duan, Jon Garibaldi, Guoping Qiu
Summary: This paper presents a new end-to-end fovea localisation method based on a hierarchical coarse-to-fine deep regression neural network, with innovative features including multi-scale feature fusion and self-attention techniques. Extensive experimental results demonstrate state-of-the-art performances, and a comprehensive ablation study and analysis validate the technical soundness and effectiveness of the overall framework and its constituent components.
IEEE TRANSACTIONS ON MEDICAL IMAGING
(2021)
Article
Engineering, Electrical & Electronic
Tianshan Liu, Kin-Man Lam, Rui Zhao, Guoping Qiu
Summary: This paper proposes a method based on cross-modal feature learning and knowledge distillation to address illumination-invariant pedestrian detection. By inserting feature learning modules at multiple levels and incorporating a segmentation auxiliary task, the multimodal network is trained end-to-end. Furthermore, a knowledge distillation framework is introduced to train a student detector with only RGB images as input, reducing the reliance on thermal images. Experimental results demonstrate the robust performance of the proposed method on a public dataset.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2022)
Article
Computer Science, Artificial Intelligence
Kanglin Liu, Gaofeng Cao, Fei Zhou, Bozhi Liu, Jiang Duan, Guoping Qiu
Summary: In this paper, a new technique called STIA-WO is presented to disentangle the latent space for unsupervised semantic face editing. By applying STIA-WO to GAN, a StyleGAN named STGAN-WO is developed, which achieves better attribute editing than state of the art methods.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2022)
Article
Geography, Physical
Qing Li, Rui Cao, Jiasong Zhu, Xianxu Hou, Jun Liu, Sen Jia, Qingquan Li, Guoping Qiu
Summary: The paper proposes a domain adaptation-based approach to improve the accuracy of indoor image localization using synthetic images. By incorporating a multi-level constrained pose regression network and a feature-level discriminator network, the proposed method effectively reduces the performance gaps between real and synthetic images.
ISPRS JOURNAL OF PHOTOGRAMMETRY AND REMOTE SENSING
(2022)
Article
Environmental Sciences
Qinglin Li, Bin Li, Jonathan M. Garibaldi, Guoping Qiu
Summary: This paper proposes a clustering-based method for representation learning of remote-sensing images. It introduces a metric to measure the discriminativeness of representations and develops an algorithm to achieve even distribution of samples while preserving their neighborhood relations.
Article
Environmental Sciences
Qinglin Li, Guoping Qiu
Summary: This paper proposes a novel method for image clustering by optimizing sample ranking and weighted training, and demonstrates its effectiveness through extensive experiments.
Article
Computer Science, Artificial Intelligence
Zhi Chen, Jiang Duan, Li Kang, Guoping Qiu
Summary: In this work, ensemble learning is embedded into deep convolutional neural networks to address the issue of class imbalance in model learning. A new loss function is designed to rectify the bias towards majority classes and improve the performance significantly.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2022)
Article
Computer Science, Artificial Intelligence
Zhi Chen, Jiang Duan, Li Kang, Guoping Qiu
Summary: This paper presents a new supervised anomaly detector called Ensemble Active Learning Generative Adversarial Network (EAL-GAN). The EAL-GAN uses a conditional GAN to generate balanced training data and introduces an innovative ensemble learning loss function and an active learning algorithm to overcome the challenges of class imbalance and high labeling cost. Extensive experimental results show that the new detector consistently outperforms other methods by significant margins.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2023)
Proceedings Paper
Computer Science, Software Engineering
Wenming Tang, Yuanhao Gong, Guoping Qiu
Summary: This paper introduces a novel algorithm for 3D mesh filtering (SAF) based on mesh structural adaptation, which achieves feature-preserving denoising by protecting corners, edges, and planes. Through experimental data, SAF has been shown to outperform or be comparable to state-of-the-art methods in feature-preserving denoising at different noise levels.
2022 IEEE 24TH INTERNATIONAL WORKSHOP ON MULTIMEDIA SIGNAL PROCESSING (MMSP)
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Cuixin Yang, Hongming Luo, Guangsen Liao, Zitao Lu, Fei Zhou, Guoping Qiu
Summary: A method using middle-layer feature loss to handle video super-resolution task was proposed, achieving superior results compared to other methods by allowing deeper network architecture.
PATTERN RECOGNITION AND COMPUTER VISION,, PT III
(2021)
Proceedings Paper
Computer Science, Artificial Intelligence
Yuanhao Gong, Wenming Tang, Lebin Zhou, Lantao Yu, Guoping Qiu
Summary: The newly proposed discrete computation scheme does not require second order differentiability, is more accurate, has a smaller support region, and is computationally more efficient.
2021 IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP)
(2021)
Article
Engineering, Electrical & Electronic
Xueyu Han, Ishtiaq Rasool Khan, Susanto Rahardja
Summary: This paper proposes a clustering-based TMO method by embedding human visual system models to adapt to different HDR scenes. The method reduces computational complexity using a hierarchical scheme for clustering and enhances local contrast by superimposing details and controlling color saturation by limiting the adaptive saturation parameter. Experimental results show that the proposed method achieves improvements in generating high quality tone-mapped images compared to competing methods.
SIGNAL PROCESSING-IMAGE COMMUNICATION
(2024)
Article
Engineering, Electrical & Electronic
Zuopeng Zhao, Tianci Zheng, Kai Hao, Junjie Xu, Shuya Cui, Xiaofeng Liu, Guangming Zhao, Jie Zhou, Chen He
Summary: The research team developed a handheld phone detection network called YOLO-PAI, which successfully achieved real-time detection and underwent testing under various conditions. Experimental results show that YOLO-PAI reduces network structure parameters and computational costs while maintaining accuracy, outperforming other popular networks in terms of speed and accuracy.
SIGNAL PROCESSING-IMAGE COMMUNICATION
(2024)
Article
Engineering, Electrical & Electronic
Vivek Sharma, Ashish Kumar Tripathi, Purva Daga, M. Nidhi, Himanshu Mittal
Summary: In this study, a novel ClGan method is proposed for automated plant disease detection. The method reduces the number of parameters and addresses the issues of vanishing gradients, training instability, and non-convergence by using an encoder-decoder network. Additionally, an improved loss function is introduced to stabilize the learning process and optimize weights effectively. Furthermore, a new plant leaf classification method called ClGanNet is introduced, achieving 99.97% training accuracy and 99.04% testing accuracy using the least number of parameters.
SIGNAL PROCESSING-IMAGE COMMUNICATION
(2024)
Article
Engineering, Electrical & Electronic
Seongeun Kim, Chang-Ock Lee
Summary: This article introduces a method for segmenting individual teeth in human teeth images by using deep neural networks to obtain pseudo edge-regions and applying active contour models for segmentation.
SIGNAL PROCESSING-IMAGE COMMUNICATION
(2024)