检索结果-内蒙古大学图书馆

FSIC: Frequency-separated image compression for small object detection

DIGITAL signal processing 2025年 156卷

作者： Dai, Chengjie Song, Tiantian Chen, Qiang Gong, Hanshen Yang, Bowei Song, Guanghua Zhejiang Univ Sch Aeronaut & Astronaut Hangzhou 310027 Peoples R China Univ Manchester Dept Math Manchester M13 9PL England Shanghai FinShine Technol Ltd Shanghai 201203 Peoples R China

The existing image compression methods are designed for the human visual system. They can achieve good compression quality for low-frequency components of the image that are important to human vision. However, for object detection models, both high and low-frequency components are essential. As a result, the detection metrics on the compressed images obtained by current methods will decline. Particularly for small object detection, the lack of high-frequency signals makes it difficult to distinguish the targets from the background. In this paper, we propose a frequency-separated image compression model, named FSIC. During the training process, the compression of low-frequency components only employs MSE loss, while the compression of high-frequency components additionally incorporates a detection loss. We validate FSIC's image compression capability for the small object detection task on the VisDrone dataset and Dota dataset. Results show that under extremely high compression rates, FSIC demonstrates a better performance compared with current compression methods. Furthermore, FSIC has the fastest encoding speed among current learning-based compression models.

关键词： Image compression Deep learning small object detection

来源：评论

学校读者我要写书评

暂无评论

Adaptive detection of radar range-Doppler dual-spread targets in lognormal-texture clutter

引用

DIGITAL signal processing 2025年 157卷

作者： Xue, Jian Fan, Zhen Xu, Shuwen Pan, Meiyan Xian Univ Posts & Telecommun Sch Commun & Informat Engn Xian 710121 Peoples R China Xidian Univ Natl Lab Radar Signal Proc Xian 710071 Peoples R China Xidian Univ Collaborat Innovat Ctr Informat Sensing & Understa Xian 710071 Peoples R China Xian Elect Engn Res Inst Xian Peoples R China

This paper investigates the problem of adaptive detection of radar targets in non-Gaussian clutter, where the target to be detected is considered to behave the dual-spread in the Doppler frequency dimension and the range dimension. The clutter is assumed to follow the compound Gaussian model with lognormal texture and unknown covariance matrix structure. The multi-rank linear subspace model and the range-spread model are employed to depict the Doppler and range spread characteristics of target echoes. Then, the range-Doppler dual-spread adaptive radar target detector with lognormal-texture is proposed using the two-step generalized likelihood ratio criteria, which replaces the true values of the unknown parameters with their maximum likelihood and maximum a posteriori estimates. Experimental results on simulated and measured data demonstrate that the proposed detector shows superior performance in different clutter and target parameters compared to the competitors.

关键词： Adaptive detection Lognormal texture Subspace model Doppler spread Range spread

来源：评论

学校读者我要写书评

暂无评论

DR-YOLO: An improved multi-scale small object detection model for drone aerial photography scenes based on YOLOv7

引用

Digital signal processing: A Review Journal 2025年 164卷

作者： Bi, Hongbo Dai, Rui Han, Fengyang Zhang, Cong Department of Electrical Information Engineering Northeast Petroleum University Daqing163000 China

With the advancement of drone technology, detecting and recognizing ground targets from aerial perspectives has become crucial in various drone applications. However, object detection in drone imagery poses several challenges, including the prevalence of small targets, the significant impact of aerial perspectives, variations in target scales, complex backgrounds, and frequent occlusions. To address these issues, we propose DR-YOLO, a multi-scale target detection model specifically designed for aerial drone images, building upon the YOLOv7 framework. We introduce the Spatial Pyramid Pooling with Dilated Convolutions (SPPDSPC) module to enhance dense target feature extraction. Additionally, we incorporate a decoupled detection head tailored for small objects and redesign the number and sizes of detection heads. To handle complex backgrounds and varying target sizes, we embed the Multi-Scale Feature Fusion (HTLF) Module into the feature pyramid network, providing rich spatial information for detection heads of different scales. Furthermore, we utilize the Gaussian Wasserstein Distance (GWD) to refine the regression loss, leading to improved bounding box quality, faster convergence, and higher accuracy in small object detection. Experimental results on the VisDrone2019 dataset demonstrate a 14.8% increase in mAP@0.5 and a 9.8% increase in mAP@0.95 compared to the baseline YOLOv7, validating the effectiveness of DR-YOLO in detecting objects within aerial drone imagery. The code and results of our method are available at https://***/DRdairuiDR/DR-YOLO. © 2025 Elsevier Inc.

关键词： Aerial photography

来源：评论

学校读者我要写书评

暂无评论

GLS: A hybrid deep learning model for radar emitter signal sorting

引用

DIGITAL signal processing 2025年 161卷

作者： Qi, Liangang Chen, Hongzhuo Guo, Qiang Huang, Shuai Kaliuzhnyi, Mykola Harbin Engn Univ Coll Informat & Commun Engn Harbin 150001 Heilongjiang Peoples R China Kharkiv Natl Univ Radio Elect Sci & Res Lab UA-61166 Kharkiv Ukraine

Radar emitter signal sorting is a pivotal aspect of radar reconnaissance signal processing. The increasing density of the electromagnetic environment in modern radar pulse streams, coupled with the growing complexity and variability of operational modes and signal forms, results in extremely limited reference data. Consequently, most existing sorting methods fall short of meeting the performance requirements of modern electronic warfare. To enhance sorting performance under conditions of limited samples and labeled data, this paper proposes a radar emitter signal sorting model based on ResGCN-BiLSTM-SE (GLS). Firstly, we propose a novel adaptive weighted adjacency matrix construction method that aggregates multi-scale information of local and global features. Based on this, for GLS networks, the graph convolutional network (ResGCN) is combined with the bidirectional long short-term memory (BiLSTM) network. The GCN is employed to extract attribute features from interleaved radar pulse sequences, while the BiLSTM is utilized to deeply capture the temporal dependence in interleaved pulse sequences after feature extraction. Finally, an improved squeeze-and-excitation (SE) module is applied to perform weighted fusion of critical channel information from both spatial and temporal features. Simulation results demonstrate that the proposed method not only achieves higher accuracy under small sample conditions compared to existing methods, but also exhibits strong robustness in challenging scenarios involving measurement errors, missing pulses, and spurious pulses.

关键词： Radar emitter signal sorting (RESS) Adaptive weighted adjacency matrix GLS model Features fusion

来源：评论

学校读者我要写书评

暂无评论

Overcoming data Scarcity in Maritime Radar Target Detection via a Complex-Valued Hybrid Spatiotemporal Network

引用

IEEE GEOSCIENCE AND REMOTE SENSING LETTERS 2025年 22卷

作者： Wang, Ju Wang, Chongyue Li, Zhaojie He, Wenjing Zhong, Yi Huang, Yan Beijing Inst Technol Sch Informat & Elect Beijing 100081 Peoples R China Chinese Acad Sci CASIA Inst Automat Beijing 100190 Peoples R China

Detecting small floating targets on the sea surface has long been a major challenge in radar signal processing. Recently, deep learning (DL) has attracted considerable attention for its potential to improve detection probability. However, its performance heavily relies on the availability of sufficiently labeled datasets, which are often difficult to acquire in complex sea clutter environments. Therefore, this letter introduces the complex-valued hybrid spatiotemporal network (CVHSTNet), a novel maritime radar target detection method designed for low-data scenarios that uses time-frequency (TF) representations of radar echoes as inputs. To mitigate the overfitting issue, CVHSTNet is intentionally designed with a shallow architecture, integrating a three-layer complex-valued convolutional neural network (CV-CNN) with a one-layer CV bidirectional long short-term memory (CV-BiLSTM) network. Unlike existing real-valued models that overlook phase information, our method operates directly on CV data to capture the complete signal representation. More importantly, this hybrid architecture enables the network to effectively exploit both spatial and temporal characteristics, thereby further enhancing feature representations. Comprehensive experiments on 40 datasets from the IPIX database demonstrate that with only 50 samples per range cell for training, the proposed method achieves a detection probability exceeding 90% in 37 of 40 datasets, with a false alarm rate (FAR) of 10(-3). To the best of our knowledge, this is the first time a DL-based approach has demonstrated the ability to distinguish between small floating targets and sea clutter under limited labeled radar data conditions.

关键词： Computer architecture Radar Clutter Feature extraction Microprocessors Radar detection Object detection Overfitting Training data mining Deep hybrid spatiotemporal network limited labeled data radar target detection sea clutter

来源：评论

学校读者我要写书评

暂无评论

Detection of small objects in remote sensing images based on bi-level routing attention and deformable convolution

引用

DIGITAL signal processing 2025年 160卷

作者： Chen, Pengbing Liu, Shouxin Feng, Wenshan Wang, Hui Li, Xiaowei Sichuan Univ Sch Elect & Informat Engn Chengdu 610065 Sichuan Peoples R China

Remote sensing images can offer detailed and abundant data regarding geographical surface features. However, existing object detection algorithms still suffer from problems such as susceptibility to complex background interference and loss of information about small objects. To address these challenges, we propose a Deformable convolution and Bi-level routing attention reinforced YOLO (DB-YOLO) algorithm. Firstly, Bi-level Routing Attention (BRA) is introduced to enhance the feature extraction capability. Following this, Deformable ConvNets v2 (DCNv2) is utilized to substitute part of traditional convolution, mitigating the loss of small object features. Finally, the Normalized Wasserstein Distance-Complete Intersection over Union (NWD-CIoU) loss, which is insensitive to the object scale, is used for localization loss. Through the experimental verification on two public remote sensing datasets, our method can improve the detection ability of small objects. Specifically, the mAP@0.5 achieves 94.8% and 97.8% on NWPU VHR-10 and RSOD, respectively.

关键词： Remote sensing images small object detection Bi-level routing attention Deformable convolution

来源：评论

学校读者我要写书评

暂无评论

Multi-view Feature Discrepancy Attack for Single Object Tracking

Multi-view Feature Discrepancy Attack for Single Object Trac...

引用

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

作者： Zhiheng Li Weng Zhimin Yuehuan Wang Huazhong University of Science and Technology Wuhan China

ISBN: (数字)9798350368741

ISBN: (纸本)9798350368758

Adversarial attacks on single object tracking (SOT) have attracted increasing attention. However, most previous works have focused on adding small digital perturbations to tracking sequences, assuming access to the data, which makes these attacks impractical in real-world applications. Inspired by traditional military camouflage, we propose a texture pattern attack method with a similar implementation, called the Multi-View Feature Discrepancy Attack (MFDA). Unlike the consistent texture features of traditional camouflage, we iteratively optimize non-planar textures by obtaining gradients from the target model to enhance feature discrepancies across different viewpoints. Moreover, the model’s multi-scale features and heatmaps are utilized as targets for our feature-level and decision-level attacks, respectively. We apply texture patterns to controllable regions of vehicle models and conduct extensive attack experiments. The results show that our method significantly degrades the performance of state-of-the-art Siamese-based trackers, and also exhibits attack capability against ViT-based trackers.

关键词： Heating systems Target tracking Perturbation methods signal processing Acoustics Surface texture Object tracking Speech processing Autonomous vehicles

来源：评论

学校读者我要写书评

暂无评论

Experimental Investigation for GPR Detection of Buried Objects with Varied Sizes and Orientation

Experimental Investigation for GPR Detection of Buried Objec...

引用

Power, Control and Computing Technologies (ICPC2T), Conference on

作者： Sujeet Roy Himanshu Govil Sandip Mukherjee Prabhat Diwan Department of Applied Geology National Institute of Technology Raipur India Independent Researcher Indian Institute of Remote Sensing Dehradun India

ISBN: (数字)9798331530402

ISBN: (纸本)9798331530419

Ground Penetration Radar (GPR) is a remote sensing technology with immense potential. Its performance and quality of results strongly depends on primarily the composition of the soil and further on other aspects of the subsurface targets like the size, material, shape, depth, and orientation. The central operating frequency of the GPR and the type of antenna on the GPR are selected based on the factors above, to be able to detect your intended target satisfactorily. Limited exploratory research work has been undertaken in this field in India towards detection of small shallow buried objects. The present study attempts to experimentally investigate the combined effect of predetermined size and orientation of small shallow buried objects in conjunction with a suitably selected central frequency of the GPR. The experiment was conducted in Jaisalmer, where the arid dry sandy wastelands are conducive to GPR performance. signal processing methods have been applied to the standard GPR results to derive better output. Apart from the standard effect of material of the targets and the depth buried to, the study mainly provided data indicating a positive correlation between the central operating frequency of the GPR and the quality of detection of the shallow subsurface targets of size corresponding to the wavelength $\lambda$. The positive correlation of polarization of the GPR antenna to the alignment of the buried object was also noted. Further investigations covering more factors affecting GPR performance is advocated by the present study.

关键词： Surveys Correlation Soil Planning data mining Test equipment Buried object detection Standards Remote sensing Antennas

来源：评论

学校读者我要写书评

暂无评论

Research on ZYNQ neural network acceleration method for aluminum surface microdefects

引用

DIGITAL signal processing 2025年 157卷

作者： Zhao, Dongxue Liu, Shenbo Zhang, Zhigang Zhang, Zhao Tang, Lijun Changsha Univ Sci & Technol Sch Phys & Elect Sci Changsha 410114 Hunan Peoples R China

Convolutional Neural Networks (CNN) are an important means of detection of microdefects on the aluminum surface, and the high complexity and computing power requirements of the CNN model lead to difficulties in deploying them on edge computing platforms as the detection accuracy continues to improve. We have studied a lightweight acceleration method for detecting microdefects on aluminum surfaces on the Zynq-7000 All Programmable SoC (ZYNQ) platform. A lightweight aluminum surface defect detection network (LADFastDet) and high-performance accelerators based on ZYNQ are designed to meet the requirements of precision and speed under limited resources. In the LADFastDet structure, a lightweight inverted residual block is designed by combining depthwise convolution, inverted residual block, and inverted bottleneck. A multiscale feature fusion structure is designed to effectively improve the detection accuracy of LADFastDet, especially small target defects. We design accelerators on ZYNQ through optimization methods such as loop optimization strategy, ping-pong buffering, and multichannel and multiple interfaces data reading and writing to reduce data access latency and thus improve the computing speed. The experimental results show that the LADFastDet model has a mAP of 97.51%, the inference time of the accelerators for a single image is 42.57 ms, and a power consumption of 2.15 W, which achieves a throughput of 24.9 GOPS and an energy efficiency of 11.58 GOPS/W.

关键词： Aluminum surface microdefects detection Lightweight detection network Hardware accelerators Fast detection of defect image signal

来源：评论

学校读者我要写书评

暂无评论

DF-CFAR: deep feature constant false alarm ratio detector based on feature game for sea-surface small target with robustness to sea states

引用

Neural Computing and Applications 2025年 1-20页

作者： Xiang, Houhong Zhu, Liangliang Chen, Yufeng Wang, Fengyu Zeng, Xiaolu School of Computer Science and Information Engineering Hefei University of Technology Hefei230601 China Department of Early Warning Technology Wuhan Early Warning Academy Wuhan430019 China Hangzhou Institute of Technology Xidian University Hangzhou311200 China School of Artificial Intelligence Beijing University of Posts and Telecommunications Beijing100876 China School of Information and Electronics Beijing Institute of Technology Beijing100081 China

The classical constant false alarm rate (CFAR) detector is optimal for target detection in Gaussian white noise but struggles with unknown, time-varying sea states. data-driven target detection methods are highly sensitive to clutter distribution, leading to poor detection performance and high false alarm rates (Pfa) under unknown sea states. This paper proposes a deep feature constant false alarm ratio (DF-CFAR) detector based on a feature game model. The input to the feature extraction network is fragmented fast time-dimensional data processed by coherent integration, incorporating a feature game mechanism. This mechanism effectively mines target echo features that are uncorrelated with the background sea clutter. By transforming the target detection problem under varying sea state conditions into a regression prediction problem, we achieve constant false alarm detection through a statistical probability threshold. Simulation and real data experiments demonstrate that, compared to commonly used algorithms, the proposed method not only achieves higher detection probability and lower false alarm rates but also exhibits excellent constant false alarm characteristics and stronger robustness to unseen sea states. Results from publicly available X-band radar data processing show that the proposed algorithm can eliminate the influence of different sea clutters and improve the detection performance of sea-surface small targets by an equivalent signal-to-clutter ratio (SCR) of about 3.1 dB. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2025.

关键词： Gaussian distribution

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：