Article
Energy & Fuels
Zhaoyun Zhang, Shihong Huang, Yanxin Li, Hui Li, Houtang Hao
Summary: This paper proposes an optical image detection method based on deep learning and morphological detection for insulator defect detection. The Faster RCNN is used to locate and extract the target image, and a segmentation method is used to remove the background. A shape transformation method for insulator detection is proposed, and a mathematical model is established to describe the defect. The experimental results show that the proposed method can accurately detect and locate insulator defects, with higher accuracy than common methods.
Article
Computer Science, Artificial Intelligence
Ce Li, Fenghua Liu, Zhiqiang Tian, Shaoyi Du, Yang Wu
Summary: This article proposes a deep learning-based salient object detection method called Dynamic and Adaptive Graph Convolutional Network (DAGCN). It utilizes an adaptive neighborhood-wise graph convolution and a spatially restricted K-nearest neighbors approach to model and analyze the context relationship in the scene. Experimental results demonstrate satisfactory performance on multiple datasets and the ability to detect camouflaged objects.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2022)
Article
Engineering, Electrical & Electronic
Qiudan Zhang, Xiaotong Xiao, Xu Wang, Shiqi Wang, Sam Kwong, Jianmin Jiang
Summary: This paper proposes a visual saliency detection method for stereoscopic images based on adaptive viewpoint feature enhancement. By analyzing the correlation between left and right views and using attention-based saliency feature pyramid extraction, the proposed method improves the accuracy of saliency detection. Additionally, a stereoscopic image saliency dataset is created to facilitate further research in this field.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2022)
Article
Engineering, Multidisciplinary
Zhuo Haoze, Han Jiaming, Zhou Guoxing, Yang Zhong
Summary: This paper proposes a detection method for insulator strings based on the InST-Net network, which uses a pretrained ResNet50 network for feature extraction and designs multiple detection branches and an SPP module to improve detection accuracy.
MATHEMATICAL PROBLEMS IN ENGINEERING
(2022)
Article
Computer Science, Information Systems
Zhen Bai, Zhi Liu, Gongyang Li, Yang Wang
Summary: In this paper, we propose a novel Adaptive Group-wise Consistency Network (AGCNet) that can adaptively adjust to image groups with random quantities to improve the performance of co-saliency detection. By introducing intra-saliency priors and an Adaptive Group-wise Consistency module, as well as specially designed decoders, our AGCNet achieves competitive performance compared to state-of-the-art models on four benchmark datasets.
IEEE TRANSACTIONS ON MULTIMEDIA
(2023)
Article
Computer Science, Information Systems
Xiaofei Zhou, Weipeng Cao, Hanxiao Gao, Zhong Ming, Jiyong Zhang
Summary: In recent years, significant progress has been made in image saliency detection, but little attention has been paid to video saliency detection. Existing video saliency models are prone to failure in challenging video scenarios, and using image saliency models directly for video saliency detection is inappropriate. To address these issues, this study proposes an end-to-end spatiotemporal integration network (STI-Net) that can effectively detect salient objects in videos. The proposed model explores spatial and temporal information comprehensively across the entire network, producing precise and complete characterization of salient objects and improving the quality of the final saliency map. Experimental results on challenging video datasets demonstrate the effectiveness of the proposed model, achieving comparable performance to state-of-the-art saliency models.
INFORMATION SCIENCES
(2023)
Article
Automation & Control Systems
Rongkai Duan, Yuhe Liao
Summary: This article proposes an improved adaptive morphological BD method for accurately extracting fault-related periodic impulses in bearing fault diagnosis. By constructing a new indicator called morphological frequency negentropy and selecting the optimal Morlet wavelet filter as the initial filter, the method's robustness is enhanced. By adaptively setting the filter length and enhancing the sampling matrix, the dependence on prior knowledge for parameter setting is reduced. Finally, the diagonal slice spectrum is used to remove residual noise. The effectiveness of the method is validated through simulation signals and real datasets, and comparison analysis with other filter methods demonstrates its superiority.
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS
(2023)
Article
Acoustics
Hao Zhou, Jianzhong Yang, Hua Xiang, Jihong Chen
Summary: The weak vibration energy in spindle motor is caused by unbalanced electromagnetic force and unqualified assembly, resulting in different fault features in different life cycles and bearing individuals. Diagnosing compound faults in the spindle motor is challenging. To solve this problem, an improved filtering and feature enhancement method combining AMF and TEO is proposed. The effectiveness of the method is verified through simulation and fault motor experiments, showing better performance in actual engineering scenarios compared to traditional methods.
Article
Computer Science, Information Systems
Bo Jiang, Xingyue Jiang, Jin Tang, Bin Luo
Summary: Co-saliency detection is an important research problem in computer vision, and this paper proposes a novel optimization framework to address this issue by integrating multiple cues for accurate estimation.
IEEE TRANSACTIONS ON MULTIMEDIA
(2021)
Article
Engineering, Electrical & Electronic
Rongkai Duan, Yuhe Liao, Shuo Wang
Summary: The adaptive morphological filter (AMF) is an improved method for analyzing faulty bearing vibration signals, which utilizes autocorrelation to enhance the morphological filter and accurately extract periodic components.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2021)
Article
Computer Science, Information Systems
Zaixing Wang, Xiaozhong Liu, Huayi Peng, Lijun Zheng, Jinhui Gao, Yufan Bao
Summary: The article proposes a railway insulator fault detection network based on convolutional neural network, which detects faulty insulators from images with high resolution and complex background. By cascading the detection network and the fault classification network, the method reduces the network calculations and improves the accuracy of fault classification.
Article
Environmental Sciences
Jiayu Xuan, Zhihui Xin, Guisheng Liao, Penghui Huang, Zhixu Wang, Yu Sun
Summary: In this paper, a SAR image-change detection method based on multiplicative fusion difference image, saliency detection, multi-scale morphological reconstruction, and fuzzy c-means clustering is proposed. The method improves the accuracy of change detection by using a new fusion DI method and saliency detection, and enhances computational efficiency by utilizing morphological reconstruction and fuzzy c-means clustering.
Article
Engineering, Electrical & Electronic
Daijie Tang, Fengrong Bi, Jiewei Lin, Xin Li, Xiao Yang, Xiaoyang Bi
Summary: This article proposes a fault detection method that combines adaptive recursive variational mode decomposition (ARVMD) and component energy distribution spectrum (CEDS) for engine fault diagnosis. The method dynamically selects the number of modes for extracting intrinsic mode functions based on energy distribution, and utilizes CEDS for fault diagnosis. Experimental results demonstrate the effectiveness and efficiency of the proposed method for engine fault diagnosis.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2022)
Article
Physics, Multidisciplinary
Wei Wang, Jianming Wang, Jianhua Chen
Summary: The paper proposes an adaptive block-based compressed video sensing scheme based on saliency detection and side information, aiming to address the issue of allocating appropriate measurement numbers for each block without the sparsity of the original signal. By fusing saliency values and significant coefficient proportions to estimate block sparsity, and introducing a global recovery model to reduce block effects in reconstructed frames, the proposed scheme achieves a significant improvement in peak signal-to-noise ratio (PSNR) compared to existing schemes at the same sampling rate.
Article
Mathematics, Applied
Cesare Bracco, Francesco Calabro, Carlotta Giannelli
Summary: We propose a method for detecting discontinuities based on null rules, computed as vectors in the null space of certain collocation matrices. These rules are used as weights to indicate the local behavior of the function and its gradient. By analyzing the properties of the rules, we introduce two indicators to distinguish between function discontinuities and gradient discontinuities. Our method is efficient and reliable, allowing for effective detection and classification of points near discontinuities.
JOURNAL OF SCIENTIFIC COMPUTING
(2023)
