检索结果-内蒙古大学图书馆

I-CenterNet: Road infrared target detection based on improved CenterNet

IET IMAGE processing 2023年第1期17卷 57-66页

作者： Li, Xiang Qian, Yurong Guo, Rui Ao, Naixiang Xinjiang Univ Sch Software Urumqi 830046 Peoples R China Key Lab Signal Detect & Proc Xinjiang Uygur Auton Urumqi Peoples R China Xinjiang Univ Software Engn Lab Urumqi Peoples R China Lianhai Chuangzhi Informat Technol Co Ltd Urumqi Xinjiang Uygur Peoples R China

Infrared target detection has strong anti-interference ability, long working distance and can work day and night. So it is widely used in military security and transportation fields, and infrared road object detection is critical in traffic checkpoints and autonomous driving. However, the target scale in infrared images changes greatly, small targets are difficult to detect, the poor image quality and low signal-to-noise ratio are still huge challenges in infrared target detection. This paper proposes an improved infrared target detection model I-CenterNet based on the anchor-free model CenterNet. The EfficientNetV2 with the channel attention mechanism is used instead of the traditional structure as the backbone network to enhance feature extraction. In order to reduce the noise of the input infrared image, Dilated-Residual U-net (DRUNet) is used. Meanwhile, feature pyramid and Sub-Pixel are combined for multi-scale feature fusion. data enhancement is implemented to improve model performance. The experimental results show that the average detection accuracy of this model on the Flir infrared data set is 87.9%, and the average detection speed reaches 14.2 frames/s.

关键词： Image recognition Computer vision and image processing techniques Military administration Traffic engineering computing

来源：评论

学校读者我要写书评

暂无评论

Dual-Level Contrast Enhancement Method for Infrared small Target Detection 2

Dual-Level Contrast Enhancement Method for Infrared Small Ta...

引用

2nd IEEE International Conference on signal, Information and data processing, ICSIDP 2024

作者： Sun, Haotian Wang, Wenzheng Xu, Yuhan Jin, Qiuyu Zhang, Zipeng Beijing Institute of Technology Beijing Key Laboratory of Embedded Real-Time Information Processing Technology Beijing China Beijing Institute of Technology School of Information and Electronics Beijing China

ISBN: (纸本)9798331515669

In the field of target detection, infrared small target detection(ISTD) is a critical area of research, playing a vital role in reconnaissance, guidance, and early warning systems. Existing spatial Filtering algorithms perform well in clean backgrounds;however, as these filters require manual design based on prior knowledge, they are prone to interference from complex scenes (e.g., clouds, buildings), leading to target omission. small targets typically exhibit blurry features and weak energy, making direct detection susceptible to interference from high-intensity areas, which may result in missed detections. To tackle these challenges, this paper presents a two-tier contrast improvement method (DLCE). First, a gradient filter group is used to enhance the signal-to-noise ratio (SNR) of the target, and then, built upon the concept of local contrast measurement (LCM), a novel intensity calculation method is developed. The new contrast window design effectively enhances the target's SNR. Compared with existing algorithms, like MPCM (Multi-Scale Patch-based Contrast Measurement) on publicly available datasets, the proposed method more effectively enhances the target's radiative intensity while suppressing background noise. The effectiveness of the proposed algorithm is confirmed by the simulation results. © 2024 IEEE.

关键词： Alarm systems

来源：评论

学校读者我要写书评

暂无评论

Semisupervised Space Target Recognition Algorithm Based on Integrated Network of Imaging and Recognition in Radar signal Domain

引用

IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS 2024年第1期60卷 506-524页

作者： Li, Chenxuan Li, Yonggang Zhu, Weigang Yang, Jun Qu, Wei He, Yonghua Space Engn Univ Grad Sch Beijing 101416 Peoples R China Space Engn Univ Dept Elect & Opt Engn Beijing 101416 Peoples R China

The application of deep learning to inverse synthetic aperture radar (ISAR) target recognition helps to improve accuracy in space target monitoring. However, the orbit transfer and maneuver of space targets are likely to cause range migration in radar echoes. Information deficiency of scattering points and the limitations of data acquisition methods pose enormous challenges to space target imaging and recognition. To address these issues, this article proposes a semisupervised space target recognition algorithm based on an integrated network of imaging and recognition in the radar signal domain. By directly processing radar signals, the algorithm can achieve high-precision space target imaging and recognition under the general situation and the conditions of range migrations. Based on the inherent characteristics of radar complex echoes, the algorithm utilizes unlabeled echoes to generate pseudolabels to achieve better generalization capabilities. Both real and complex convolutions are exploited to generate high-resolution features. Besides, feature optimization modules are designed to effectively integrate high-resolution texture features and contour features to magnify the difference between the target and background. The ablation experiments and contrast experiments indicate that, under the conditions of migration and missing components, the algorithm can obtain high-resolution features for imaging and recognition by only using a small amount of labeled data. Thus, the algorithm achieves high accuracy and robustness for migrated space target recognition and has superiority over other algorithms.

关键词： Radar imaging Space vehicles Radar Imaging Target recognition Spaceborne radar Image recognition

来源：评论

学校读者我要写书评

暂无评论

Multiband Radar signal Fusion and Extrapolation Method Based on Transformer Model 2

Multiband Radar Signal Fusion and Extrapolation Method Based...

引用

2nd IEEE International Conference on signal, Information and data processing, ICSIDP 2024

作者： Gong, Chunxu Li, Weidong Lu, Ruiming Wang, Rui Beijing Institute of Technology Radar Technology Research Institute Beijng China

ISBN: (纸本)9798331515669

The improvement of range resolution is crucial for enhancing the capability of imaging radars to detect small targets, such as drones. However, due to technical and cost limitations, current radars cannot indefinitely increase hardware bandwidth to enhance range resolution. To address this challenge, the multiband fusion and extrapolation technique has emerged. This method uses multiple discrete sub-band radar signals and fuses them to extrapolate the missing frequency bands, effectively achieving a wideband radar signal. Traditional approaches primarily rely on constructing signal models and estimating model parameters to accomplish fusion and extrapolation. This paper is the first to apply machine learning techniques to the field of multiband fusion and extrapolation, proposing a radar multiband signal fusion and extrapolation method based on a Transformer model. By training the Transformer network, the method enables accurate estimation of unknown frequency band signals. The effectiveness of the approach is validated through electromagnetic simulation data of drone echoes, demonstrating significantly improved estimation accuracy compared to traditional signal fusion and extrapolation methods. © 2024 IEEE.

关键词： Drones

来源：评论

学校读者我要写书评

暂无评论

Target Location in Distributed MIMO radar based on Coherent processing 2

Target Location in Distributed MIMO radar based on Coherent ...

引用

2nd IEEE International Conference on signal, Information and data processing, ICSIDP 2024

作者： Wei, Fa Yan, Shi Zhu, Shengqi Lan, Lan Li, Ximin Qi, Jiawei Ma, Runlong Xidian University National Key Laboratory of Radar Signal Processing Xi'an710071 China Beijing Institute of Remote Sensing Device Beijing100854 China

ISBN: (纸本)9798331515669

In this paper, the problem of target location is investigated in the distributed multiple-input multiple-output (MIMO) radar based on coherent processing. At the modelling stage, the distributed MIMO radar is divided into three separated subarrays, where both transmit and receive functions are used in subarrays 1 and 3, while only receive function is adopted in subarray 2. Furthermore, a small frequency increment is introduced among different antenna elements in the subarray 1, in the meantime, a phase modulating operation is performed for various antenna elements in the subarray 3. In this respect, the target location performance can be improved significantly by means of the transmit modulation in subarrays 1 and 3. Under the circumstances, the model of target location is established, and a two-step estimation technology is shown, where a small location sector with respect to targets are firstly obtained, and the accurate location are further estimated. Specifically, in order to reduce the computational cost required by the multi-dimensional search procedures, an alternating procedure is developed by a sequence of 1-D searches obtained alternating between the optimization over each variable keeping the other fixed. Numerical results corroborate the effectiveness of the developed methods for the target location with distributed MIMO radar based on coherent processing. © 2024 IEEE.

关键词： Phase modulation

来源：评论

学校读者我要写书评

暂无评论

Infrared small target detection based on long-short term frame difference algorithm(LSTFD) 6

Infrared small target detection based on long-short term fra...

引用

6th International Conference on Intelligent Control, Measurement and signal processing, ICMSP 2024

作者： Zhu, Bing Shi, Jiahui Yu, Ruixing Gao, Weixin Wang, Siyu Xi'an Shiyou University School of Electronic Engineering Xi'an China Northwestern Polytechnical University School of Astronautics Xi'an China

ISBN: (纸本)9798331516550

According to the characteristic that the position change of small infrared target in sequence images is more obvious than background and noise, an infrared small target detection algorithm based on multi-frame time domain information is proposed, called the Long-Short Term Frame Difference (LSTFD) algorithm. The algorithm first adopts the SURF (Speeded Up Robust Feature) alignment strategy to eliminate the relative motion of background pixels caused by camera shake and achieve global motion compensation. Then, the image sequence is constructed based on the current frame, and the computational results of long-interval difference and short-interval difference are combined to detect the infrared small targets. The experimental results show that the LSTFD algorithm can accurately detect moving targets with high detection accuracy and low false alarm rate under the condition of low signal-to-noise ratio. And the average detection rate can reach over 90% on five sets of experimental data. © 2024 IEEE.

关键词： Motion compensation

来源：评论

学校读者我要写书评

暂无评论

Underwater Sonar Target Detection Based on Improved ScEMA-YOLOv8

引用

IEEE GEOSCIENCE AND REMOTE SENSING LETTERS 2024年 21卷 1页

作者： Zheng, Linhan Hu, Tao Zhu, Jin Jiangsu Univ Sci & Technol Ocean Coll Zhenjiang 212013 Peoples R China Key Lab Zhenjiang Smart Ocean Informat Percept & T Zhenjiang 212003 Peoples R China

Underwater target detection is mainly achieved through two methods: optical imaging and sonar scanning. Compared to optical imaging, sonar target detection has the characteristics of strong penetration and long scanning distance, which makes it more suitable for tasks such as deep sea, turbid water, and long-distance target detection. However, currently, sonar image detection still faces the following challenges: 1) difficulty in obtaining underwater sonar images and scarcity of existing open-source sonar datasets;2) the quality of sonar image is poor, which is limited by environmental noise interference, sonar equipment and related signal processing technology;3) compared to optical images, sonar images are more difficult to detect small targets;and 4) due to the influence of underwater terrain, debris, and the degree of self decay. There are significant differences in the distribution of targets in sonar images, and different types of sonar (such as side scan sonar, forward view sonar, and so on) have significant differences in visual presentation. Therefore, our letter proposes an underwater target detection framework based on improved ScEMA-YOLOv8 and conducts comparative experiments on data enhancement and transfer learning. Experimental results have shown that the improved model achieves 98.4% and 97.6% mAP@0.5 and it can also achieve high precision and meet the requirements of real-time.

关键词： Sonar Sonar detection Feature extraction data models Training Oceans Transfer learning ScEMA-YOLOv8 side-scan sonar sonar target detection transfer learning

来源：评论

学校读者我要写书评

暂无评论

GASS: GENERALIZING AUDIO SOURCE SEPARATION WITH LARGE-SCALE data 49

GASS: GENERALIZING AUDIO SOURCE SEPARATION WITH LARGE-SCALE ...

引用

49th IEEE International Conference on Acoustics, Speech, and signal processing (ICASSP)

作者： Pons, Jordi Liu, Xiaoyu Pascual, Santiago Serra, Joan Dolby Labs San Francisco CA 94103 USA

ISBN: (纸本)9798350344868;9798350344851

Universal source separation targets at separating the audio sources of an arbitrary mix, removing the constraint to operate on a specific domain like speech or music. Yet, the potential of universal source separation is limited because most existing works focus on mixes with predominantly sound events, and small training datasets also limit its potential for supervised learning. Here, we study a single general audio source separation (GASS) model trained to separate speech, music, and sound events in a supervised fashion with a large-scale dataset. We assess GASS models on a diverse set of tasks. Our strong in-distribution results show the feasibility of GASS models, and the competitive out-of-distribution performance in sound event and speech separation shows its generalization abilities. Yet, it is challenging for GASS models to generalize for separating out-of-distribution cinematic and music content. We also fine-tune GASS models on each dataset and consistently outperform the ones without pre-training. All fine-tuned models (except the music separation one) obtain state-of-the-art results in their respective benchmarks.

关键词： General audio source separation deep learning

来源：评论

学校读者我要写书评

暂无评论

Active Underwater Target Detection Using a Shallow Neural Network With Spectrogram-Based Temporal Variation Features

引用

IEEE JOURNAL OF OCEANIC ENGINEERING 2024年第1期49卷 279-293页

作者： Choo, Youngmin Lee, Keunhwa Hong, Wooyoung Byun, Sung-Hoon Yang, Haesang Sejong Univ Dept Def Syst Engn Seoul 05006 South Korea Korea Res Inst Ships & Ocean Engn KRISO Ocean Syst Engn Res Div Daejeon 34103 South Korea Seoul Natl Univ Dept Naval Architecture & Ocean Engn Seoul 08826 South Korea

In this article, we propose an active target detector by using a shallow neural network (NN) with novel features under small sonar data, where deep learning (DL) models are restricted. The features are contrived by considering differences between target and clutter signals in the temporal variation of amplitude and frequency when using an active ping;these are extracted from preprocessed spectrograms of beam time series. The classification ability of the features is demonstrated through comparison with that of prevalent timbral features in the training data set that reflect the human auditory system. Finally, the detection performance of the shallow NN trained using the proposed features is compared with those from the fine-tuned DL models. During the generalization test, the DL models miss a significant number of targets in the test data set-including target signals with patterns unseen during training-and the VGG16 used for detailed comparison has a detection probability of 0.69. The probability increases to 0.86 via the shallow NN, with remarkably enhanced target detection. This demonstrates the competitive performance and generalization ability of the shallow NN, using features that consider active sonar returns under a lack of data.

关键词： Sonar Clutter Feature extraction Detectors Sonar detection Spectrogram Object detection Automatic target detection feature analysis machine learning (ML) sonar signal processing supervised learning

来源：评论

学校读者我要写书评

暂无评论

Research and Implementation of Radar signal processing Algorithm for FPGA-Based DBF System

Research and Implementation of Radar Signal Processing Algor...

引用

2023 IEEE MTT-S International Microwave Workshop Series on Advanced Materials and Processes for RF and THz Applications, IMWS-AMP 2023

作者： Chen, Shuqiong Xu, Wenwen Xu, Dalong Wang, Hao Cao, Jianyin Nanjing University of Science and Technology 200 Xiaolingwei Street Nanjing210094 China

ISBN: (纸本)9798350308761

In order to address the issue of low energy echoes from 'low, slow, small' targets, this paper investigates a signal processing system with FPGA as the core architecture. The real-time signal processing for 64 channels and wideband-narrowband pulse echoes are achieved in this paper. Besides, DBF is utilized in the elevation dimension, providing advantages such as a small blind zone, the ability to detect and track targets at various distances, as well as simultaneous formation of multiple beams. This real-time radar processing system, compared to traditional FPGA+DSP architecture phased-array radars, offers advantages of high-speed data transmission, high parallelism, high stability, and high reliability. © 2023 IEEE.

关键词：

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：