检索结果-内蒙古大学图书馆

real-time Aerial image Edge Detection with MicroWatt Power ISA on FPGA 3

Real-Time Aerial Image Edge Detection with MicroWatt Power I...

3rd International conference on Intelligent and Innovative Technologies in Computing, Electrical and Electronics, IITCEE 2025

作者： Joshi, Prashant V. Sudharshan, K.M. Asuti, Majunath G. Poornashree, S.J. Research, Sunitha K. School of Ece Reva University Bengaluru India Bnmit Department of Ece Bengaluru India

ISBN: (纸本)9798331515911

This research investigates the implementation of real-time aerial image edge detection using the Canny edge detection algorithm with the MicroWatt Power Instruction Set Architecture (ISA)-Open Core Processor on Field Programmable Gate Arrays (FPGAs). Canny edge detection is particularly advantageous for aerial image processing due to its ability to produce precise and well-defined edges, making it ideal for applications in surveillance, agriculture, and environmental monitoring where detail preservation is critical. The proposed system utilizes the MicroWatt Power ISA-Open core processor to optimize resource utilization, achieving high throughput and low latency while maintaining energy efficiency. The work uses the Xilinx Vivado EDA tool to calculate the power dissipation of the FPGA implementation, demonstrating significant reductions in power consumption compared to traditional processing methods. The experimental results showcase the effectiveness of the Canny edge detector in producing accurate edge maps while minimizing resource usage, thus enabling real-time processing capabilities. This work contributes to the advancement of low-power embedded systems for aerial imaging, highlighting the potential of FPGAs combined with the MicroWatt Power ISA for efficient and effective image analysis. © 2025 IEEE.

关键词： Aerial photography

来源：评论

学校读者我要写书评

暂无评论

Evaluation of Transfer Learning for Visual Multiple Target Tracking 1

Evaluation of Transfer Learning for Visual Multiple Target T...

引用

1st International conference on Advances in Computer Science, Electrical, Electronics, and Communication Technologies, CE2CT 2025

作者： Jadhav, Anisha Anil California State University Department of Computer Science Monterey Bay Seaside United States

ISBN: (纸本)9798331518578

Deep learning has revolutionized high-level image processing tasks, notably image classification and segmentation, by effectively handling multi-dimensional features in image space. This report investigates the application of transfer learning, utilizing features extracted from a pre-trained VGG deep neural network to visual multiple-target tracking (MTT). Traditional feature detectors like the Lucas-Kanade Tracker and Good Features to Track (GFTT) algorithm are computationally efficient but limited in dynamic, cluttered environments, such as those encountered by unmanned aerial systems (UAS). Replacing GFTT with deep feature detectors from VGG is proposed to assess tracking accuracy and improve computational efficiency. Our experiments reveal that, though promising, deep feature extraction results in lower frame rates and less smooth trajectories than traditional methods. This study highlights the potential and current limitations of integrating deep learning features into real-time visual tracking systems. Plans include optimizing the deep learning model for real-time processing, automating filter selection for feature extraction, and exploring additional pre-trained networks to improve tracking performance and efficiency. © 2025 IEEE.

关键词： image segmentation

来源：评论

学校读者我要写书评

暂无评论

Hyperspectral and Multispectral image Fusion for Remotely Sensed Target Detection: A New Cloud-Edge Collaborative Approach

引用

IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING 2025年 63卷

作者： Wu, Zebin Liu, Chenxin Sun, Jin Hu, Jun Wei, Zhihui Plaza, Javier Plaza, Antonio Nanjing Univ Sci & Technol Sch Comp Sci & Engn Nanjing 210094 Jiangsu Peoples R China Univ Extremadura Dept Technol Comp & Commun Hyperspectral Comp Lab Caceres 10071 Spain

Hyperspectral target detection (HTD) can provide detailed information about the objects and materials within a scene and holds significant importance in remote sensing image analysis. Traditional HTD methods often suffer from low detection accuracy when applied to hyperspectral images (HSIs) with low spatial resolution, where the target only occupies a few pixels. This can be addressed by exploiting the higher spatial resolution of multispectral images (MSIs). In addition, many cloud-based HTD methods, which rely on the distributed processing capability of cloud computing to cope with large-scale datasets, may result in long transmission delays that cannot meet real-time requirements. This article suggests a cloud-edge collaborative HTD approach based on the fusion of remotely sensed HSIs and MSIs. We first introduce an HTD algorithm that employs low-rank matrix decomposition and hierarchical constraint energy minimization (hCEM) to fuse a low-resolution HSI (LR-HIS) and a high-resolution MSI (HR-MSI). Aiming at a continuous shooting scenario, we further present a cloud-edge implementation of the HTD algorithm through the collaboration of a cloud cluster and edge servers deployed close to data acquisition devices. The overall processing flow of remotely sensed data fusion in the cloud-edge environment is formulated as a flowshop scheduling-like optimization problem. We develop a co-optimization scheduling algorithm to explore the best resource allocation solutions to the formulated problem. Experimental results on both general-purpose and real-world datasets show that the newly proposed HTD algorithm leads to significant improvements in detection accuracy over traditional methods, and the cloud-edge collaborative approach further enhances computational efficiency.

关键词： Cloud computing Matrix decomposition Collaboration Spatial resolution Data integration Servers Detectors Object detection Hyperspectral imaging Clustering algorithms Cloud-edge collaborative computing hyperspectral target detection (HTD) multiobjective optimization

来源：评论

学校读者我要写书评

暂无评论

Toward Comprehensive Semantic Prompt for Region Contrastive Learning Underwater image Enhancement

Toward Comprehensive Semantic Prompt for Region Contrastive ...

引用

2025 IEEE International conference on Acoustics, Speech, and Signal processing, ICASSP 2025

作者： Wang, Xiao Fu, Yongsheng Wang, Wei Liu, Wei School of Computer Science and Technology Wuhan University of Science and Technology Wuhan China Hubei Province Key Laboratory of Intelligent Information Processing and Real-time Industrial System China

ISBN: (纸本)9798350368741

Underwater image enhancement (UIE) focuses on mitigating image quality degradation due to light absorption and scattering. However, most existing methods enhance images via a global and uniform manner, neglecting the inherent semantic information in different regions, which may cause the network to easily deviate from the region's original color. Moreover, these methods typically rely on clear images to guide network convergence, a process constrained by the limited availability of real-world datasets, making it extremely challenging to train enhancement models for various degradations. To address these challenges, this paper introduces a semantic guidance and region contrastive constraints network (SRCNet). Initially, we propose a semantic-aware RWKV (Receptance Weighted Key Value) block and a semantic prompt regularization module. These components leverage intra-target semantic correlations to preserve image details and colors within a global perceptual framework, while employing focal loss to emphasize the restoration of severely degraded regions. Subsequently, we introduce a region contrastive learning method that effectively utilizes negative samples to precisely capture features sensitive to degradation factors, thereby fostering robust feature distributions. Finally, experimental results demonstrate that our method outperforms existing state-of-the-art (SOTA) approaches. © 2025 IEEE.

关键词： contrastive learning rwkv semantic prompt Underwater image enhancement

来源：评论

学校读者我要写书评

暂无评论

VAGeo: View-specific Attention for Cross-View Object Geo-Localization

VAGeo: View-specific Attention for Cross-View Object Geo-Loc...

引用

2025 IEEE International conference on Acoustics, Speech, and Signal processing, ICASSP 2025

作者： Li, Zhongyang Yuan, Xin Liu, Wei Xu, Xin School of Computer Science and Technology Wuhan University of Science and Technology China Hubei Province Key Laboratory of Intelligent Information Processing and Real-Time Industrial System China

ISBN: (纸本)9798350368741

Cross-view object geo-localization (CVOGL) aims to locate an object of interest in a captured ground- or drone-view image within the satellite image. However, existing works treat ground-view and drone-view query images equivalently, overlooking their inherent viewpoint discrepancies and the spatial correlation between the query image and the satellite-view reference image. To this end, this paper proposes a novel View-specific Attention Geo-localization method (VAGeo) for accurate CVOGL. Specifically, VAGeo contains two key modules: view-specific positional encoding (VSPE) module and channel-spatial hybrid attention (CSHA) module. In object-level, according to the characteristics of different viewpoints of ground and drone query images, viewpoint-specific positional codings are designed to more accurately identify the click-point object of the query image in the VSPE module. In feature-level, a hybrid attention in the CSHA module is introduced by combining channel attention and spatial attention mechanisms simultaneously for learning discriminative features. Extensive experimental results demonstrate that the proposed VAGeo gains a significant performance improvement, i.e., improving acc@0.25/acc@0.5 on the CVOGL dataset from 45.43%/42.24% to 48.21%/45.22% for ground-view, and from 61.97%/57.66% to 66.19%/61.87% for drone-view. © 2025 IEEE.

关键词： Cross-view Object-level Localization Hybrid Attention Mechanism View-specific Positional Encoding

来源：评论

学校读者我要写书评

暂无评论

Research on design of image processing and recognition based on deep learning

Research on design of image processing and recognition based...

引用

2024 International conference on Physics, Photonics, and Optical Engineering, ICPPOE 2024

作者： Li, Yanan Wang, Hua Li, Yongbo Guangdong University of Science and Technology Guangdong Dongguan523668 China

ISBN: (数字)9781510689121

ISBN: (纸本)9781510689114

To achieve target detection and defect recognition in power inspection images, an image processing and recognition algorithm based on deep learning is proposed. This algorithm mainly adopts an improved Faster-RCNN model, combined with HSI color feature extraction, for real-time image processing and identification of various conventional faults. Firstly, the collected power inspection images are normalized, converting the RGB colors of the images to HSI colors. By traversing each pixel in HSI space, it is determined whether there is a fault in the pixel based on the color characteristics of the image. Secondly, a DenseNet network is constructed and connected to the RoIAlign layer and prediction layer. Finally, the identification of target defects is achieved through Faster-RCNN. Simulation results show that the algorithm can accurately identify fault defects with a precision of up to 94.60%, indicating a broad application prospect in power equipment defect recognition. © 2025 SPIE.

关键词： RGB color model

来源：评论

学校读者我要写书评

暂无评论

Face Recognition Attendance Framework using Quantum Edge processing 3

Face Recognition Attendance Framework using Quantum Edge Pro...

引用

3rd International conference on Intelligent and Innovative Technologies in Computing, Electrical and Electronics, IITCEE 2025

作者： Rajarajeswari, S. Ranjitha, Hassan Srinivas Ramaiah Institute of Technology Bangalore India

ISBN: (纸本)9798331515911

Through the use of a Streamlit web application, the script is a highly intelligent tool that combines cutting-edge quantum processing principles with traditional computer vision approaches to handle a variety of face detection and identification tasks. While other libraries handle computer vision jobs and numerical calculations, such as NumPy and OpenCV, Streamlit is essential to developing the web application interface. While Streamlit WebRTC allows real-time video streaming, PIL is utilized for image processing. The application is set up to have a personalized look and a centered layout. Users may upload photographs, choose detection modes, and modify face detection settings on its header and sidebar. The program detects faces in uploaded photographs by encircling them with bounding boxes using the Haar Cascade Classifier. This may be used to build a face dataset or train a face recognition algorithm. Another important feature is real-time face identification, in which the software uses the Haar Cascade Classifier to identify faces in a video stream. Bounding boxes are used to emphasize faces, and photos containing faces are retained for later study or training. The program may utilize taken photographs to train a model for face recognition. These photos are processed, turned into grayscale, and then the LBPH (Local Binary Patterns Histograms) technique is used to train the model. The trained model is then kept for use in upcoming facial recognition assignments. Moreover, the program may be set up to track attendance in real time by identifying faces in a webcam stream. time stamps are recorded, attendance is logged using recognized faces, and the information is saved to a CSV file. All things considered, the script provides a complete solution that combines traditional and quantum methods for real-time face identification, picture processing, and attendance monitoring. This makes it an adaptable tool for a range of computer vision and machine learning application

关键词： Photography

来源：评论

学校读者我要写书评

暂无评论

Hybrid Quantum-Classical Transfer Learning for real-time Data processing 17

Hybrid Quantum-Classical Transfer Learning for Real-Time Dat...

引用

17th International conference on COMmunication Systems and NETworkS, COMSNETS 2025

作者： Chittem, Mahesh Babu Kumar Pradhan, Ashok SRM University AP Dept of Computer Science and Engineering Amaravati India

ISBN: (纸本)9798331531195

Transfer learning is a set of techniques to apply skills or knowledge from a source task to a target task that is different but related, while Hybrid Quantum-Classical Transfer Learning (HQCTL) model extends the skills learned with quantum feature extraction specifically for edge computing which lacks resources. HQCTL combined with quantum-derived characteristics enhances accuracy, time, and real-time computation when it comes to classical aspects such as object detection or image analysis. In experimenting with images datasets such as COCO and PASCAL VOC the distribution of the framework generally presented higher accuracy and lower costs in terms of computation compared to either a purely classical or quantum approach. Of course, quantum enhanced feature extraction is still far from known and has greater potential for HQCTL as it helps to advance data representation which is optimal for the strict real-time processing in the IoT periphery. Possible research avenues include the development of different quantum representations of the problem, enhancements of the approach interconnectivity with various edge substrates, and the application of the framework to new machine learning tasks such as video analysis and time series prediction. Through presenting the HQCTL framework the potential of hybrid quantum-classical models to enhance edge AI applications while offering reliability, scalabilty and efficiency is demonstrated. © 2025 IEEE.

关键词： Transfer learning

来源：评论

学校读者我要写书评

暂无评论

Automated Blood Group Detection System using image processing for Non- Invasive Antigen Feature Extraction and Classification 3

Automated Blood Group Detection System using Image Processin...

引用

3rd International conference on Intelligent Data Communication Technologies and Internet of Things, IDCIoT 2025

作者： Reddy, Gopu Chandrasekhar SaiKrishna, Gajavada Annie Micheal, A. Department of CSE with AIML Sathyabama Institute of Science and Technology Chennai India Department of CSE Sathyabama Institute of Science and Technology Chennai India

ISBN: (纸本)9798331527549

Blood group detection is essential in medical diagnostics and transfusion medicine but often relies on invasive methods requiring specialized infrastructure. This research introduces an automated, non-invasive blood group detection system using YOLOv8 object detection and advanced image processing. The system addresses challenges such as human error, lack of accessibility in resource-limited settings, and delays in traditional methods. By analyzing high-resolution blood sample images, the system extracts antigen features and classifies agglutination patterns for A, B, and D blood groups. With a mean average precision (mAP) of 97.4% and an inference time of 15 milliseconds per image, the system delivers scalable, real-time, and robust diagnostics. This research advances healthcare accessibility and precision medicine objectives, offering an efficient and practical alternative to traditional laboratory-based methods." © 2025 IEEE.

关键词： Antigens

来源：评论

学校读者我要写书评

暂无评论

An Efficient image-Based Virtual Scene Generation and Stitching Algorithm

An Efficient Image-Based Virtual Scene Generation and Stitch...

引用

2025 International conference on Multi-Agent Systems for Collaborative Intelligence, ICMSCI 2025

作者： Wei, Luo Chongqing Three Gorges Vocational College Wanzhou Chongqing400155 China

ISBN: (纸本)9798331509828

With the development of computer graphics and image processing, the virtual scene generation and splicing technology has been widely used in various fields of computer-aided image analysis. In this context, this study proposes a new and efficient image stitching algorithm to solve the image stitching artifacts caused by perspective changes and exposure differences in multi-camera systems, and expands it to the field of virtual images. First, as a pre-processing part, the model analyzes the image through nonlinear filtering and mesh deformation optimization to ensure image quality. Subsequently, the model presents a new image stitching algorithm based on the image matching method of minimizing projection deviation and thin plate spline theory. This algorithm further proposes an improved EIEOWF algorithm based on circular weighted fusion to achieve the task of smooth transition stitching of images. Finally, the stitching technology was extended to virtual scene generation, and OpenGL technology was used to achieve real-time 3D texture mapping. The test part is verified through experiments on the UDIS-D data set. The results show that the average splicing accuracy of the proposed algorithm in 100 sets of tests is 95%, with a standard deviation of 2%, which is significantly better than the existing comparison algorithm. © 2025 IEEE.

关键词： image matching

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：