Edge Extraction Research Articles

EGSDK-Net: Edge-Guided Stepwise Dual Kernel Update Network for Panoptic Segmentation

In recent years, panoptic segmentation has garnered increasing attention from researchers aiming to better understand scenes in images. Although many excellent studies have been proposed, they share some common unresolved issues. Firstly, panoptic segmentation, as a novel task, is still confined within inherent frameworks. Secondly, the prevalent kernel update strategies do not adequately utilize the information from each stage. To address these two issues, redwe propose an edge-guided stepwise dual kernel update network (EGSDK-Net) for panoptic segmentation; the core components are the real-time edge guidance module and the stepwise dual kernel update module. The first component, after extracting and positioning edge features through an extra branch, applies these features to the normally transmitted feature maps within the network to highlight the edges. The input image is initially processed with the Canny edge detector to generate and store the predicted edge map, which acts as the ground truth for supervising the extracted edge feature map. The stepwise dual kernel update module enhances the utilization of information by allowing each stage to update both its own kernel and that of the subsequent stage, thereby improving the judgment capabilities of the kernels. redEGSDK-Net achieves a PQ of 60.6, representing a 2.19% improvement over RT-K-Net.

MITD-Net: Multi-scale iterative tamper detection network for image manipulation localization

Advancements in image tampering technology have escalated social risks and security challenges associated with tampered images. Accurately identifying tampered regions remains a significant challenge. To address the issue effectively, a multi-scale iterative tampering detection network is proposed, which decomposes the task into two phases, the Global-Local Feature Synchronization Phase and the Tampered Area Refinement Phase. Initially, dual-branch preprocessing and parallel feature extraction are employed. The Enhanced Downsampling Attention Block (EDAB) is introduced, and an Edge Enhancement Module (EEM) is implemented to extract edge features. Subsequently, Multi-Scale Dilated Convolution (MSDC) is utilized in the tampered area refinement phase, along with the proposed Pixel-level Feature Clustering Module (PFCM) and an iterative mechanism. Finally, incorporating a cross-scale fusion mechanism enables the synthesis of feature information from various scales, thereby enhancing the model's ability to detect subtle alterations in tampered regions. MITD-Net was validated on publicly available datasets, including CASIA, Columbia, COVERAGE, NIST 16, and IMD 20, achieving AUC scores of 84.5%, 98.5%, 85.8%, 86.6%, and 82.6%, respectively. Numerous experiments indicate that MITD-Net achieves higher detection accuracy, exhibits superior robustness, and can adapt to various tampering types and styles.

Phase contrast ghost imaging using pseudo-thermal light

Ghost imaging and ghost holography enable non-local imaging of amplitude and phase objects. Recent studies have indicated that dual-arm phase-contrast imaging can be achieved by extracting only the edges of the target, thereby obviating the need for a full image reconstruction. However, existing approaches still depend on either cameras or complex scanning procedures. In this paper, we combine spiral phase contrast imaging with ghost imaging and propose a computational dual-arm phase contrast imaging scheme based on Fourier basis modulation under speckle illumination, with single-point detection used in both arms. Experimental and simulation results validate the feasibility of the proposed scheme and demonstrate its effective implementation under compressed sampling. Furthermore, we also experimentally investigate the impact of the lateral coherence length of the speckle on edge extraction accuracy, as well as the effect of the number of speckles on the imaging signal-to-noise ratio.

Multifunctional Artificial Electric Synapse of MoSe2-Based Memristor toward Neuromorphic Application.

Research on memristive devices to seamlessly integrate and replicate the dynamic behaviors of biological synapses will illuminate the mechanisms underlying parallel processing and information storage in the human brain, thereby affording novel insights for the advancement of artificial intelligence. Here, an artificial electric synapse is demonstrated on a one-step Mo-selenized MoSe2 memristor, having not only long-term stable resistive switching characteristics (reset 0.51 ± 0.01 V, on/off ratio > 30, retention > 103 s) but also diverse electrically adjustable synaptic behaviors, including multilevel conductance (synaptic weight), excitatory postsynaptic current (EPSC), paired-pulse facilitation (PPF), long-term potentiation/depression (LTP/D), spike-timing-dependent plasticity (STDP), and especially activity-dependent synaptic plasticity (ADSP). More significantly, neuromorphic functions of both image edge extraction and biological perception imitation have been successfully achieved. These results present a promising design toward synaptic devices for advancing neuromorphic systems with integrated brain-like neural sensing, memory, and recognition.

DCFE-YOLO: A novel fabric defect detection method.

Accurate detection of fabric defects is crucial for quality control in the textile industry. However, the task of fabric defect detection remains highly challenging due to the complex textures and diverse defect patterns. To address the issues of inaccurate localization and false positives caused by complex textures and varying defect sizes, this paper proposes an improved YOLOv8-based fabric defect detection method. First, Dynamic Snake Convolution is introduced into the backbone network to enhance sensitivity to elongated and subtle defects, improving the extraction of edge and texture details. Second, a Channel Priority Convolutional Attention mechanism is incorporated after the Spatial Pyramid Pooling layer to enable more precise defect localization by leveraging multi-scale structures and channel priors. Finally, the feature fusion network integrates Partial Convolution and Efficient Multi-scale Attention, optimizing the fusion of information across different feature levels and spatial scales, which enhances the richness and accuracy of feature representations while reducing computational complexity. Experimental results demonstrate a significant improvement in detection performance. Specifically, mAP@0.5 increased by 2.9%, precision improved by 3.5%, and mAP@0.5:0.95 rose by 2.3%, highlighting the model's superior capability in detecting complex defects. The project is available at https://github.com/lilian998/fabric.

A Face Fatigue Detection Model for Edge Information Extraction

In contemporary society, fatigue driving is a major cause of traffic accidents, making accurate and timely detection critical for improving driving safety. In this study, we propose a novel fatigue detection model, CSA-YOLO, designed to enhance the accuracy and efficiency of facial fatigue detection. The model is based on the YOLOv9s network model and introduces several key improvements to address the limitations of traditional methods, which often lose critical edge information. First, the Cross-Stage Partial Network (C3 module) replaces the RepNCSPELAN4 module to enhance the model’s ability to extract edge information effectively. Second, the incorporation of the SimAM attention mechanism improves feature selection, while the Content-Aware ReAssembly of Features (CARAFE) upsampling operator enhances the quality of reconstructed image details. Experimental results demonstrate that the proposed CSA-YOLO model achieves significant performance improvements, with a 2.24% increase in accuracy, a 2.58% improvement in recall, and a 2.44% boost in mAP compared to the original YOLOv9s model. These results highlight the model’s potential for practical application in reducing the risks of fatigue-related accidents.

A New Approach to COVID-19 Detection Using Alexnet From Chest CT Images

Background: At the beginning of 2020, the pneumonia-like COVID-19 virus spread rapidly worldwide. With the emergence of this dangerous virus, work, and daily life have become very difficult. To control this virus, all the centers are closed and quarantined by the government and the countries of the world. Every day, many people from all over the world die due to COVID-19. Many efforts are being made in all fields to diagnose and treat people infected with this virus. For this reason, many researchers started working on identifying this virus and its types. The Scientifics, in computer science, did not sit idle either. Method: In some studies, different image processing methods and algorithms have been used to extract the edge features of the Computed Tomography (CT) images of the lung. In this clinical-computerized study, lung CT images of non-infected and infected people from different hospitals in Lorestan province were used. This collection has 90 stereotypes of images in jpg format of CT scans of people's lungs, each file has more than 200 images from different angles of lung imaging. The preprocessing methods were the first step of the research. In the second phase, edge detection was applied to the dataset to get the highest accuracy rate. Consequently, a classification Convolutional Neural Network (CNN)-AlexNet architecture was used to reach the final aim. Results: The results show that the average accuracy rate of image edge extraction with a threshold value of 0.1 is 93% and the accuracy rate of AlexNet architecture classification is 100%. The proposed method helps physicians to improve disease diagnosis from lung CT images to achieve a more accurate detection rate. Conclusion: This study shows that the CNN-AlexNet architecture effectively increases the diagnosis accuracy rate than the other methods. It is suggested that educational programs for researchers in the field of disease detection from radiology images be provided and that the effectiveness of different types of deep learning methods be compared in future studies.

A Self-Supervised Pretraining Framework for Context-Aware Building Edge Extraction From 3-D Point Clouds

A Self-Supervised Pretraining Framework for Context-Aware Building Edge Extraction From 3-D Point Clouds

Optimising source code vulnerability detection using deep learning and deep graph network

To enhance the effectiveness of vulnerability detection in software developed using C and C++ programming languages, our study introduces a novel correlation calculation method for analyzing and evaluating Code Property Graphs (CPG). The intelligent computation method proposed in this study comprises three key stages. In the first stage, we present a method for extracting features from the CPG source code. To accomplish this, we integrate three distinct data exploration methods: employing Graph Convolutional Neural (GCN) to extract node features from CPG, utilizing Convolutional Neural Network (CNN) to extract edge features from CPG, and finally employing the Doc2vec natural language processing algorithm to extract source code from CPG nodes. The second stage involves proposing a method for synthesizing CPG source code features. Building on the features acquired in the first stage, our paper introduces a synthesis and construction method to generate feature vectors for the source code. The final stage, stage three, executes the detection of source code vulnerabilities. The experimental results demonstrate that our proposed model in this study achieves higher efficiency compared to other studies, with an improvement ranging from 3% to 4%.

Research on Intelligent Identification Method for Pantograph Positioning and Skateboard Structural Anomalies Based on Improved YOLO v8 Algorithm

The abnormal structural state of the pantograph skateboard is a significant and highly concerning issue that has a significant impact on the safety of high-speed railway operation. In order to obtain real-time information on the abnormal state of the skateboard in advance, an intelligent defect identification model suitable to be used as a monitoring device for the pantograph skateboard was designed using a computer vision-based intelligent detection technology for pantograph skateboard defects, combined with an improved YOLO v8 model and traditional image processing algorithms such as edge extraction. The results show that the anomaly detection algorithm for the pantograph sliding plate structure has good robustness, maintaining recognition accuracy of 90% or above in complex scenes, and the average runtime is 12.32 ms. Railway field experiments have proven that the intelligent recognition model meets the actual detection requirements of railway sites and has strong practical application value.

An unsupervised underwater image enhancement method based on generative adversarial networks with edge extraction

An unsupervised underwater image enhancement method based on generative adversarial networks with edge extraction

A novel OL-mapping operator-based edge detection approach

Accurate edge detection is of great significance to image processing. However, the traditional methods, which use differential operations or prior experience to determine the value of the operator, cannot meet the high demand for accuracy in the field of image processing. Deep-learning based detection algorithms restrict its flexible usability due to the high demand of computational resources (storage space, processor) for its large number of parameters. To date, neither traditional nor emerging techniques cannot achieve SOTA performance on detection tasks. Therefore, to address the need for both precision and resources, this paper proposes an edge detection operator (HQ operator) based on the relationship between the original image and label mapping, which enhances the edge detection effect while drastically reducing the computational complexity. Unlike previous edge extraction operators, HQ operator utilizes a combination of labels and original images for edge detection, and does not use iterative training, but through subgraph generation classification (SGC) method and subgraph feature fusion operation (SFFO) to obtain the edge detection operator at once. Compared to the traditional edge detection operator, HQ operator can accurately extract feature information of the edge of target region. And compared to deep learning, the number of parameters of the HQ operator is greatly reduced and the calculation time of parameters is reduced. In addition, the HQ operator obtained on one dataset can be ported to other datasets with good edge extraction results. The edge images processed by the HQ operator can be spliced with the original image in the form of channels, and the image after splicing is fed into the image segmentation network to improve the performance of the segmentation network. to improve the performance of the segmentation network. The experimental results show that by using the HQ operator, the image edge detection effect is significantly improved (PA:96.49%), and the performance of the segmentation network is also improved to a certain extent when this operator is combined with the segmentation network(Dice:89.97%, IoU:81.97%).

Object Detection Based on Edge Enhanced YOLOv5

Abstract In order to address issues that arise in complex scenarios, an edge-enhanced YOLOv5 algorithm is proposed. The fusion of edge features with original feature maps serves to enhance the accuracy of the detection process. The algorithm employs Sobel and Canny operators for edge extraction, integrating these with low-level feature maps through addition and concatenation. Testing in urban and desert settings demonstrated that the enhanced models exhibited superior accuracy and a reduction in false and missed detection compared to the original YOLOv5. This methodology has the potential to enhance target detection performance in complex environments.

Ferroelectric‐Switchable Single Photodetector Implementing Complex Optoelectronic Logics

AbstractOptoelectronic logic gates (OELGs) are promising building blocks of edge extraction, image recognition, and data encryption in logic circuits. However, implementing complex OELGs based on CMOS devices needs additional circuit configuration due to the fixed photodetection mode, which inevitably makes the manufacturing process of chip complicated and large. This work proposes a ferroelectric‐switchable photodetector (FeS‐PD) in 2D α‐In2Se3/WSe2 heterojunction, which can implement the complex OELG functions. The FeS‐PD exhibits the ferroelectric‐tunability between the photovoltaic and photoconductive mode via the sign and amplitude of polarization voltage, as well as accompanied by a high current photo/dark ratio. Moreover, the non‐volatile‐tunability of photovoltaic voltage (opening circuit voltage) distinctively enables the bias one input of logics, which enables a single device to implement the complex XOR and AND logic functions via programming gate pulse voltage, light input, and bias. Compared to the traditional CMOS‐based logic gate, FeS‐PD significantly reduces the transistor number by 91.7% and 50% for nonlinear XOR and linear AND logics, respectively. The research provides a simple platform for on‐chip optoelectronic‐logic computing circuits.

DStokes-CGCP: A low-light color polarization image enhancement method combining chroma spectrum and global contour awareness

In view of the challenges posed by low light environments, we propose a method for enhancing color polarization images, called DStokes-CGCP. This method employs a dual-branch architecture to effectively leverage the distinct characteristics of different components of the Stokes vector, thereby enhancing the utilization of polarization information. Firstly, the chroma spectrum branch (f(CS)) extracts features in the spatial and frequency domains based on the low-frequency information of S0, effectively fusing the information between feature maps of different dimensions, thereby compensating for the loss of spatial domain information and improving image clarity. Secondly, the global contour awareness branch (f(GCA)) captures the details and edge information of the image through edge and global feature extraction for the high-frequency components of S1 and S2, and further enhances the clarity and texture details of S1 and S2. Additionally, this study introduces a new dataset, named RCPI. Experimental results show that DStokes-CGCP performs well in improving the quality of low-light images and enhancing the naturalness of the effect.

Visual Edge Feature Detection and Guidance under 3D Interference: A Case Study on Deep Groove Edge Features for Manufacturing Robots with 3D Vision Sensors

For manufacturing robots equipped with 3D vision sensors, the presence of environmental interference significantly impedes the precision of edge extraction. Existing edge feature extraction methods often enhance adaptability to interference at the expense of final extraction precision. This paper introduces a novel 3D visual edge detection method that ensures greater precision while maintaining adaptability, capable of addressing various forms of interference in real manufacturing scenarios. To address the challenge, data-driven and traditional visual approaches are integrated. Deep groove edge feature extraction and guidance tasks are used as a case study. R-CNN and improved OTSU algorithm with adaptive threshold are combined to identify groove features. Subsequently, a scale adaptive average slope sliding window algorithm is devised to extract groove edge points, along with a corresponding continuity evaluation algorithm. Real data is used to validate the performance of the proposed method. The experiment results show that the average error in processing interfered data is 0.29mm, with an average maximum error of 0.54mm, exhibiting superior overall performance and precision compared to traditional and data-driven methods.

Detection of motor nervous disease using deep learning based Duple feature extraction network

Background A motor nervous disease (MND) is a debilitating nervous disease that affects motor neurons that regulates the muscular voluntary movements. The disease gradually destroys parts of the neurological system. Generally, MND develops owing to a grouping of genetic, behavioural, and natural features. Objective However, early detection of MND is challenging and manual identification requires a lot of time. Therefore, automated methods like deep learning structures are needed to detect MND quickly and more accurately than manual classification. In this work, a novel deep learning-based Duple feature extraction network is proposed for identifying MND in its early stages. Methods Initially, the input DTI images are pre-processed utilizing a Gaussian adaptive bilateral filter (GAB) to improve the quality of the image. Then the pre-processed DTI images are fed into the dual feature extraction phase for colour and structural conversion. The Colour Information Feature (CIF) with Local and Global sampling (LOG) is integrated into the LinkNet module to extract colour features. Moreover, the Local Binary Pattern (LBP) with Edge sampling models is integrated into the MobileNet module to extract edge features. Afterward, the extracted colour and texture features of images are flattered and given as the input to a Deep Neural Network for classifying the MND levels. Results From the test results, the proposed Duple feature extraction network has yielded a 99.62% accuracy rate. The proposed DNN improves its F1-score by 1.32%, 2.1%, and 3.18% better than FNN, GNN, and GRU respectively. The proposed Duple-feature extraction network improves overall accuracy by 6.15%, 5.56%, 5.96%, and 6.68% compared to CNN, SVM-RFE, MLP, and Tri-planar CNN respectively. Conclusion The novel deep learning-based Duple feature extraction framework shows promising results in early detection of motor nervous disease, significantly improving accuracy and f1-scores compared to existing models.

Lightweight UAV Small Target Detection and Perception Based on Improved YOLOv8-E

Traditional unmanned aerial vehicle (UAV) detection methods struggle with multi-scale variations during flight, complex backgrounds, and low accuracy, whereas existing deep learning detection methods have high accuracy but high dependence on equipment, making it difficult to detect small UAV targets efficiently. To address the above challenges, this paper proposes an improved lightweight high-precision model, YOLOv8-E (Enhanced YOLOv8), for the fast and accurate detection and identification of small UAVs in complex environments. First, a Sobel filter is introduced to enhance the C2f module to form the C2f-ESCFFM (Edge-Sensitive Cross-Stage Feature Fusion Module) module, which achieves higher computational efficiency and feature representation capacity while preserving detection accuracy as much as possible by fusing the SobelConv branch for edge extraction and the convolution branch to extract spatial information. Second, the neck network is based on the HSFPN (High-level Screening-feature Pyramid Network) architecture, and the CAA (Context Anchor Attention) mechanism is introduced to enhance the semantic parsing of low-level features to form a new CAHS-FPN (Context-Augmented Hierarchical Scale Feature Pyramid Network) network, enabling the fusion of deep and shallow features. This improves the feature representation capability of the model, allowing it to detect targets of different sizes efficiently. Finally, the optimized detail-enhanced convolution (DEConv) technique is introduced into the head network, forming the LSCOD (Lightweight Shared Convolutional Object Detector Head) module, enhancing the generalization ability of the model by integrating a priori information and adopting the strategy of shared convolution. This ensures that the model enhances its localization and classification performance without increasing parameters or computational costs, thus effectively improving the detection performance of small UAV targets. The experimental results show that compared with the baseline model, the YOLOv8-E model achieved (mean average precision at IoU = 0.5) an mAP@0.5 improvement of 6.3%, reaching 98.4%, whereas the model parameter scale was reduced by more than 50%. Overall, YOLOv8-E significantly reduces the demand for computational resources while ensuring high-precision detection.

An online open circuit faults diagnosis method for converter using the lightweight two-channel deep network

Online diagnostic technology is crucial for ensuring the reliable operation of power converters in industrial applications. However, quickly diagnosing the faults of a power converter using intelligent algorithms is challenging due to the redundancy and noise of the sensor measurement signal. Hence, this paper proposes a lightweight two-channel deep network (LTCDN) to overcome it and achieve rapid fault diagnosis. It consists of a quadratic mean preprocessing layer (QMPL), a two-channel convolutional layer (TCCL), and a bidirectional gated recurrent unit (BGRU). The QMPL is designed to achieve denoising and lightening of output currents. Then, the TCCL is proposed to extract edge features and internal key information for improving diagnostic accuracy. Finally, the BGRU classifier completes the online fault diagnosis with a fast detection speed and high precision. It is evaluated through experiments with real-time open circuit faults for multiple switches of a four-level active neutral point clamped (4L-ANPC) converter, demonstrating superior performance with a false diagnosis rate of less than 1% and a diagnosis time of under 8ms. The method combines a novel data preprocessing layer with an efficient neural network to provide an effective fault detection method for a 4L-ANPC converter, which not only has high accuracy but also has significant computational efficiency. Its scalability for more complex multilevel converters is also verified.

Extraction of asymptotic edges of microcracks in silicon nitride bearings based on adaptive nonlocal mean filtering and iterative tracking algorithm

The gradual change edge of the Si3N4 bearing roller microcracks with decreasing gray gradients are difficult to be extracted by threshold segmentation. The method for extracting the gradual change edge of Si3N4 bearing roller microcracks using adaptive non-local mean filtering and an iterative tracking algorithm is proposed. Widely distributed, large-span, dense noise is eliminated from the gradual change edges of microcracks in Si3N4 bearing roller microcrack images. In the iterative expansion process of microcrack defect shape, the gradual change edge pixel of microcrack is accurately tracked. The Si3N4 bearing roller microcrack image is enhanced by adaptive non-local mean filtering. The following are the experimental findings: The PSNR and SNR reach 38.12 dB and 40.94 dB, respectively. The microcrack gradual change edge pixels can be extracted with an edge coverage rate of 92.5 % and an accuracy of 93.8 % using the iterative tracking algorithm for Si3N4 bearing roller microcrack images.