检索结果-内蒙古大学图书馆

Efficient multi-level cross-modal fusion and detection network for infrared and visible image

ALEXANDRIA ENGINEERING JOURNAL 2024年 108卷 306-318页

作者： Gao, Hongwei Wang, Yutong Sun, Jian Jiang, Yueqiu Gai, Yonggang Yu, Jiahui Shenyang Ligong Univ Sch Automat & Elect Engn Shenyang 110159 Peoples R China Chinese Acad Sci Shenyang Inst Automat State Key Lab Robot Shenyang 110016 Peoples R China Zhejiang Univ Dept Biomed Engn Hangzhou 310027 Peoples R China Binjiang Inst Zhejiang Univ Innovat Ctr Smart Med Technol & Devices Hangzhou 310053 Peoples R China

With the rapid development of uncrewed aerial vehicle (UAV) technology, detecting aerial images has found significant applications across various domains. However, existing algorithms overlook the impact of illumination on target detection, resulting in less satisfactory detection performance under low-light conditions. We propose EfficientFuseDet, a visible and infrared image fusion detection network to overcome this issue. First, an effective multilevel cross-modal fusion network called EfficientFuse is presented to combine complementary information from both modalities better. EfficientFuse captures local dependencies and global contextual information in shallow and deep layers, seamlessly combining complimentary local and global features throughout the network. The generated fused images can exhibit clear target contours and abundant texture information. Second, we propose a detection network called AFI-YOLO, which employs an inverted residual vision transformer backbone (IRViT) to effectively address the challenges associated with background interference in fused images. We design an efficient feature pyramid network (EFPN) that efficiently integrates multiscale information, enhancing multiscale detection capability using aerial images. A reparameterization decoupling head (RepHead) is proposed to further improve target classification and localization precision. Finally, experiments on the DroneVehicle dataset indicate that the detection accuracy using fused images can reach 47.2%, which is higher than that observed with visible light images of 45 %. Compared to state-of-the-art detection algorithms, EfficientFuseDet exhibits a slight decrease in speed. However, it demonstrates superior detection capabilities and effectively enhances the detection accuracy using aerial images under low-light conditions.

关键词： Uncrewed aerial vehicles Aerial image Image fusion Object detection

来源：评论

学校读者我要写书评

暂无评论

Obstacle detection Method of Underground Electric Locomotive Rail Based on Instance Segmentation

引用

TRANSPORTATION RESEARCH RECORD 2024年第6期2678卷 708-723页

作者： Tong, Jiale Wang, Shuang Guo, Yongcun Wang, Wenshan Yang, Tun Zong, Shuqi Anhui Univ Sci & Technol Sch Mech Engn Huainan Peoples R China State Key Lab Min Response & Disaster Prevent & Co Huainan Peoples R China Collaborat Innovat Ctr Min Intelligent Technol & E Huainan Peoples R China

Real-time and accurate obstacle detection is a vital technology for electric locomotives, especially as driverless vehicles are introduced. A method of obstacle detection for underground electric locomotive rail based on instance segmentation is developed to solve the problems of misdetection and missing detection, low detection accuracy, and slow detection speed of rail obstacles. The method of locating the track mask, demarcating the effective driving boundary, expanding the track mask, and forming the effective driving area is adopted to verify whether the target is an obstacle based on whether the target is located in the effective driving area, to avoid the problem of misdetection and missing detection of the target obstacle. The YOLACT++ (You Only Look At CoefficienTs) model is improved, and path augmentation and target classification loss function replacement strategies are adopted to enhance the model's ability to detect target details and increase the accuracy of target segmentation. Compared with traditional image processing, this method can detect both straight rail and turnout. The mean average precision of boundary box mAP0.5(box) and mask mAP0.5(mask) of the improved YOLACT++ model reaches 98.52% and 98.55%, which is higher than that of the YOLACT++ model, and the detection frame rate reaches 21.9 frames per second.

关键词： rail rail safety computer models system safety train

来源：评论

学校读者我要写书评

暂无评论

Simulation of infrared Radiation Characteristics and Dynamic Imaging of Space targets Based on Spaceborne detection

Simulation of Infrared Radiation Characteristics and Dynamic...

引用

2024 Applied Optics and Photonics China

作者： Yan, Lining Cao, Yisen Cao, Yunhua Xidian Univ Sch Phys 2 South Taibai Rd Xian Shaanxi Peoples R China

ISBN: (数字)9781510687769

ISBN: (纸本)9781510687752;9781510687769

The infrared radiation characteristics of space targets have been applied in the fields of target detection, classification, and identification. This paper simulates the infrared radiation characteristics and dynamic imaging of space targets using spaceborne detection systems. The study investigates the infrared radiation characteristics of space targets in the 4.2 similar to 6 mu m, 5.5 similar to 7 mu m, and 8 similar to 12 mu m bands. Additionally, a spaceborne network detection scenario was established to update the relative positions of detection satellites and targets in real-time. By modeling and simulating the infrared radiation characteristics of mid-course warheads and decoys, a sequence of satellite-detected infrared images under various detection conditions was generated. This research provides a reference for infrared imaging methods for the detection, classification, and identification of space targets.

关键词： Space targets infrared radiation characteristics infrared imaging Spaceborne detection systems

来源：评论

学校读者我要写书评

暂无评论

ACDF-YOLO: Attentive and Cross-Differential Fusion Network for Multimodal Remote Sensing Object detection

引用

REMOTE SENSING 2024年第18期16卷 3532页

作者： Fei, Xuan Guo, Mengyao Li, Yan Yu, Renping Sun, Le Henan Univ Technol Key Lab Grain Informat Proc & Control Minist Educ Zhengzhou 450001 Peoples R China Henan Univ Technol Sch Artificial Intelligence & Big Data Zhengzhou 450001 Peoples R China Zhengzhou Univ Sch Elect & Informat Engn Zhengzhou 450001 Peoples R China Nanjing Univ Informat Sci & Technol Jiangsu Collaborat Innovat Ctr Atmospher Environm Nanjing 210044 Peoples R China

Object detection in remote sensing images has received significant attention for a wide range of applications. However, traditional unimodal remote sensing images, whether based on visible light or infrared, have limitations that cannot be ignored. Visible light images are susceptible to ambient lighting conditions, and their detection accuracy can be greatly reduced. infrared images often lack rich texture information, resulting in a high false-detection rate during target identification and classification. To address these challenges, we propose a novel multimodal fusion network detection model, named ACDF-YOLO, basedon the lightweight and efficient YOLOv5 structure, which aims to amalgamate synergistic data from both visible and infrared imagery, thereby enhancing the efficiency of target identification in remote sensing imagery. Firstly, a novel efficient shuffle attention module is designed to assist in extracting the features of various modalities. Secondly, deeper multimodal information fusion is achieved by introducing a new cross-modal difference module to fuse the features that have been acquired. Finally, we combine the two modules mentioned above in an effective manner to achieve ACDF. The ACDF not only enhances the characterization ability for the fused features but also further refines the capture and reinforcement of important channel features. Experimental validation was performed using several publicly available multimodal real-world and remote sensing datasets. Compared with other advanced unimodal and multimodal methods, ACDF-YOLO separately achieved a 95.87% and 78.10% mAP0.5 on the LLVIP and VEDAI datasets, demonstrating that the deep fusion of different modal information can effectively improve the accuracy of object detection.

关键词： multimodal fusion attention cross-modal difference module remote sensing images object detection

来源：评论

学校读者我要写书评

暂无评论

Geothermal target detection integrating multi-source and multi-temporal thermal infrared data

引用

ORE GEOLOGY REVIEWS 2024年 167卷

作者： Chao, Jiangqin Zhao, Zhifang Xu, Shiguang Lai, Zhibin Liu, Jianyu Zhao, Fei Yang, Haiying Chen, Qi Yunnan Univ Inst Int Rivers & Ecosecur Kunming 650500 Peoples R China Zhaotong Univ Sch Geog Sci & Tourism Zhaotong 657000 Peoples R China Yunnan Univ Sch Earth Sci Kunming 650500 Peoples R China Yunnan Key Lab Sanjiang Metallogeny & Resources Ex Kunming 650500 Peoples R China MNR Key Lab Sanjiang Metallogeny & Resources Explorat Kunming 650500 Peoples R China Yunnan Int Joint Lab China Laos Bangladesh Myanmar Kunming 650500 Peoples R China Yunnan Geol Engn Explorat Grp Co Ltd Kunming 650041 Peoples R China

Thermal infrared remote sensing (TIRS) technology is intriguing for geothermal anomaly detection due to its time efficiency and cost-effectiveness. Land surface temperature (LST) derived from TIRS indicates potential geothermal anomalies. However, LST on different natural features, exhibiting relatively cold/hot anomalies in daytime and nighttime, affected the identification of abnormal areas. The aim of this study was to highlight geothermal anomalous areas by integrating multi -view daytime and nighttime LST data. Specifically, the winter LST time series of drillings on 2013 - 2023 was processed based on Landsat-8 TIRS in diurnal scenarios. An information aggregation classification method about Dempster -Shafer evidence theory was presented to target recognition at nighttime by the ASTER LST product. This day -night information fusion analysis effectively highlighted potential areas with geothermal anomalies. The results demonstrated that the geothermal anomalies were mainly distributed along the Ruili-Longling Fault and Wanding Fault in a Northeast (NE) - Southwest (SW) direction, and the NE side was more significant than the SW side. These findings suggested that the east side was closer to the magma heat source, and the presumed magma heat source location aligned closely with the detected geothermal anomalies. Geological data were employed for geological interpretation of the LST anomaly area, eliminating non -geothermal influences. Along with drilling data verification, this method was successfully used to identify geothermal anomaly areas in Ruili City, Yunnan Province. Overall, this study provided valuable insights into the detection of geothermal anomalies through TIRS, contributing to the successful development of geothermal resources.

关键词： Geothermal anomalies Thermal infrared remote sensing Daytime and Nighttime land surface temperature Dempster-Shafer evidence theory Time series Ruili city

来源：评论

学校读者我要写书评

暂无评论

detection and recognition of UA targets with multiple sensors

引用

AERONAUTICAL JOURNAL 2023年第1308期127卷 167-192页

作者： Chen, W. S. Chen, X. L. Liu, J. Wang, Q. B. Lu, X. F. Huang, Y. F. China Acad Civil Aviat Sci & Technol Beijing 100028 Peoples R China Naval Aviat Univ Yantai 264001 Peoples R China Beihang Univ Beijing 100191 Peoples R China

Modern low-altitude unmanned aircraft (UA) detection and surveillance systems mostly adopt the multi-sensor fusion technology scheme of radar, visible light, infrared, acoustic and radio detection. Firstly, this paper summarises the latest research progress of UA and bird target detection and recognition technology based on radar, and provides an effective way of detection and recognition from the aspects of echo modeling and micro motion characteristic cognition, manoeuver feature enhancement and extraction, motion trajectory difference, deep learning intelligent classification, etc. Furthermore, this paper also analyses the target feature extraction and recognition algorithms represented by deep learning for other kinds of sensor data. Finally, after a comparison of the detection ability of various detection technologies, a technical scheme for low-altitude UA surveillance system based on four types of sensors is proposed, with a detailed description of its main performance indicators.

关键词： UA detection Deep learning Multi-sensor Information fusion

来源：评论

学校读者我要写书评

暂无评论

infrared Ship detection Based on Improved YOLO-11 16

Infrared Ship Detection Based on Improved YOLO-11

引用

16th IEEE International Conference on Computational Intelligence and Communication Networks, CICN 2024

作者： Hu, Yayuan School of Computing Northeast Electric Power University Jilin China

ISBN: (纸本)9798331505264

In the marine field, infrared detection technology is of great significance for timely localization and detection of ships in security missions. However, since infrared ship targets are often in the environmental conditions of small pixel occupancy, low contrast and complex background, it poses a great challenge for multiship detection, classification, and localization tasks. Therefore, to solve these problems, we propose a lightweight enhanced small target infrared ship target detection (YOLO-FFE) network, Designed lightweight and efficient C2FFA module to increase the feature extraction capability of the backbone network. A lightweight attention mechanism, ELA, is added to the small-target detection header to better learn and predict the locations and bounding boxes of small targets in ship images. Experiments demonstrate that the proposed YOLO-FFE has better detection performance for infrared ship targets. © 2024 IEEE.

关键词： Image enhancement

来源：评论

学校读者我要写书评

暂无评论

An enhance ranging algorithm based on multi-waveform classification with hyperspectral LiDAR

引用

MEASUREMENT 2025年 253卷

作者： Li, Xingyun Shao, Hui Lu, Yuan Chen, Yuwei Sun, Long Dai, Hui Hong, Tao Zhang, Xinyuan Zhang, Hulong Anhui Jianzhu Univ Anhui Int Joint Res Ctr Ancient Architecture Intel Hefei 230601 Peoples R China Infrared & Low Temp Plasma Key Lab Anhui Prov Hefei 230037 Peoples R China Adv Laser Technol Lab Anhui Prov Hefei 230037 Peoples R China Anhui Jianzhu Univ Anhui Prov Engn Res Ctr Reg Environm Hlth & Spatia Hefei 230601 Peoples R China

When the light detection and ranging (LiDAR) detected system interrogates targets with high retro-reflective properties, the backscattered pulse energy is prone to exceed the maximum receiving range of the photodetector. It will cause signal saturation because the system cannot record the complete echo waveform, which leads to the deviation of ranging results. To solve the problem, this study proposes an enhanced ranging algorithm based on multi-waveform classification (RMWC) with 101-band hyperspectral LiDAR (HSL). Considering the difference of echo waveforms collected in different bands, the threshold method is combined with random forest classifier to classify echo waveforms into normal waveform, saturated waveform and U-shaped waveform. Then peak value method (PV), centroid method (CM) and waveform fitting method (WF) are selected to calculate the corresponding time of flight of the three waveforms respectively. Finally, the target distance is determined by the weighted average of the mode values of the bands' ranging results. Compared with PV, CM and WF ranging results at 905 nm, the experimental results verify that the proposed algorithm can effectively reduce the ranging error caused by waveform saturation. For planar retro-reflective target, the optimal average error (AE) and standard deviation (Std) are 0.0053 m and 0.0083 m, respectively. For diffuse reflection whiteboard, the AE is 0.0010 m and the Std is 0.0053 m. The reconstruction results of planar target point cloud show that RMWC ranging results are better than these traditional single-band ranging methods, which can provide a reference for the optimal design of high precision laser ranging system.

关键词： Laser ranging Saturated waveform Multi-waveform classification Waveform time discrimination

来源：评论

学校读者我要写书评

暂无评论

Imaging detection and classification for Near-space Dim targets 7

Imaging Detection and Classification for Near-space Dim Targ...

引用

7th International Conference on Vision, Image and Signal Processing, ICVISP 2023

作者： Yuan, Yupu Xi, Runping Chen, Qian School of Computer Science Northwestern Polytechnical University Xi'an China Shaanxi Province Key Laboratory of Speech & Image Processing Xi'an China National Engineering Laboratory of Integrated Aero-Space-Ground-Ocean Big Data Application Technology Xi'an China

ISBN: (纸本)9781839539848

The motion mode of near-space targets is complex due to their high threat level. The target imaging faces low SNR and susceptibility to background noise. The existing detection and classification algorithms struggle to meet the demands of performance and speed on such targets. To address these issues, a YOLOv5-based target detection algorithm utilizes CARAFE upsampling and the small detection head to extract internal features and edge information of dim target. The algorithm adeptly captures dim target features, enhancing real-time performance and detection accuracy while overcoming subtle feature granularity and extraction difficulties. Aiming at the problem that there is no existing trajectory classification algorithm in the near-space field, we propose an infrared dim target trajectory classification algorithm. The improved attention module is used to enhance the extraction of motion features, and the target category is accurately identified by the prediction module, improving the capacity to extract dependencies between trajectory fragments. We execute comparative experiments on common infrared dim target dataset, near-space infrared dim target dataset and real dataset to confirm the viability of our methods. The outcomes indicate that our proposed method can be applied to the above three types of dim targets and significantly outperforms state-of-the-art methods. © The Institution of Engineering & technology 2023.

关键词： Extraction

来源：评论

学校读者我要写书评

暂无评论

Micro Solder Defect Inspection Using infrared Sequence and Deep Learning Learning Omni-scale CNN 24

Micro Solder Defect Inspection Using Infrared Sequence and D...

引用

24th International Conference on Electronic Packaging technology (ICEPT)

作者： Jiang, Ye Liu, Zhiyong Liao, Guanglan Huazhong Univ Sci & Technol Sch Mech Sci & Engn Wuhan Peoples R China

ISBN: (纸本)9798350338812;9798350338829

The chip on PCBA is developing towards miniaturization and high density. It becomes more difficult to detect micro solder bump defects inside the electronic package and soldering defects cause 71% of electronic package problems. However, fault diagnosis in the industrial manufacturing scenario is more complex and diverse involving the detection and location of solder bump defects at the IC level. As an attractive option to package inspection, X-ray may seriously affect the service life of ICs. Our group proposed a nondestructive diagnosis method on the basis of active thermography. In the recorded thermal image, the solder ball was segmented for the classification analysis. Due to the large size of PCBA, the narrow gap laser is limited to industrial applications. Because of the inefficiency and low identification rate of current detection, here we propose a detection method using infrared thermal imaging and Omni-scale CNN deep learning algorithm. We take the real chip as the experimental target. The infrared thermal sequence is collected to analyze the temperature evolution of different defect types. We use the target region segmentation to extract features to distinguish the invisible solder defects in infrared images. With the hyperparameter optimization, the Omni-scale CNN classification model realizes efficient and highly accurate detection of different solder (400 mu m-500 mu m) defect types.

关键词： Solder bump infrared sequence Chip on board Defect detection Deep learning

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：