Search

With the rapid development of deep learning, convolutional neural networks have achieved milestones in synthetic aperture radar (SAR) image object detection. However, object detection in SAR images is still a great challenge due to the difficulty in distinguishing targets from complex backgrounds. At the same time, most of the targets in SAR images are small and unevenly distributed, which makes it challenging to extract sufficient feature information. To solve these issues mentioned above, an efficient object detection network for SAR images based on Swin transformer and YOLOv7 is proposed in this article. First, we design a novel feature aggregation module Petty-specialized feature aggregation (PS-FPN) to enrich small targets’ semantic and spatial features while keeping the model lightweight. PS-FPN module uses the fusion of deep and shallow features by using cross-layer feature aggregation and single-branch feature aggregation to enhance the detection of small targets. Second, a novel attention mechanism strategy mix-attention is proposed to find more attention regions. Finally, we add one more prediction head to extract shallow features that effectively preserve small targets’ feature information. To verify the effectiveness of the proposed algorithm, extensive experiments are carried out on several challenging SAR image datasets. The results show that, compared with other state-of-the-art detectors, the proposed method can achieve significant performance based on lightweight detection.

A light field (LF) camera captures both spatial and angular information of the real world, and the intertwined nature of these dimensions presents a pressing challenge in effectively disentangling meaningful LF information for depth estimation. This paper introduces a feature extraction network based on LF decoupling, which ingeniously separates the LF. Furthermore, given the extensive volume of input data inherent in LF images, a novel reparameterizable Residual-Densely Branched Leaky-ReLU Block(Res-DBLB) architecture was developed to replace conventional residual structures and multibranch architectures and enhance inference efficiency. Incorporating an attention mechanism further refines the network, effectively addressing the computational intensity and time-consuming nature of LF depth estimation, thus furthering the advancement of this technology. Our model was applied to widely used datasets as well as the latest LF datasets, namely HCI and UrbanLF, showing superior performance over six other popular models across four evaluation metrics.

The protection of electromagnetic pulse (EMP) is important for aircraft due to vulnerable electronic equipments are installed. The control box of engine is also suffering from EMP and seldom studied. Meanwhile, most of work related to EMP protection focus on shielding design. Few works consider the effect of engine shell. The field distribution induced by EMP and its influences on the controller layout are unaddressed. This work will comprehensively investigate the field distribution characteristics of an engine under EMP, which is studied by simulating and experimenting methods. The transmission line matrix (TLM) is used to simulated field distribution of the bounded wave simulator and aero-engine, which provides better placement for equipment under test in the simulator. The field differences of bounded wave simulator and plane wave have been analyzed by using CST software. The feasibility of plane excitation instead of bounded wave simulator model excitation is checked, which has been verified by using the engine model. Then, the distribution characteristics of transverse and longitudinal fields surrounding the aero-engine are obtained, and the surface current distribution of the engine is also obtained in the simulation. The aero-engine simulation is excited by bounded wave simulator under CST software. According to the corresponding monitoring points, the simulated field distribution characteristics are verified by using the largely bounded wave simulator. Finally, the influence of field distribution on the layout of the engine electronic control system is analyzed, which provides a reference for the anti-electromagnetic pulse design of the engine electromagnetic controller.

Visibility restoration of an image degraded by scattering media (such as haze, fog, smog) is a challenging task which requires careful design of computational methods. The fundamental problem which limits clear image reconstruction is the random diffusion of light in the scattering medium. Recently, the dark channel prior method has emerged as a powerful approach for haze removal under adverse weather conditions. However, this method requires high computational time, and fails to achieve haze-free image in the sky region. To overcome these problems, we propose a novel imaging method based on quadtree theory combined with transmission optimization (QTCTO) to restore the visibility of images degraded by scattering medium. The proposed method uses the quadtree theory (for region segmentation) to estimate the global atmospheric light, while the transmission in the media is estimated by coefficient modification for the dark channel. The experimental results show that the proposed method is robust against scattering (under smoke and turbid water conditions) and has the advantage of low computational time. The proposed method can be used for security surveillance, remote sensing, and various military applications.

Inspired by human vision, retina-like imaging has the feature of rotation and scaling invariance, which is beneficial for target tracking and recognition. However, there is a limitation that the center of field of view (FOV) should match with the target centroid. Currently, there is no effective method to quantitatively estimate the effects from this mismatch. To solve the issue, a novel evaluation method is proposed by applying the log-polar transform and Fourier transform. Then reconstruct the target image by exchanging magnitude and phase information between the target image and its rotated or scaled counterpart. Using two current methods of digital image processing, scale-invariant feature transform (SIFT) and biaxial projection similarity analysis (BPS), as evaluation criteria, the proposed method performs efficiently for all eccentric angles. Simulation results show that for a specific target, the invariance of rotation and scaling works effectively for a value above 0.69.

Convolutional neural networks (CNN) with their deep feature extraction capability have recently been applied in numerous image fusion tasks. However, the image fusion of infrared and visible images leads to loss of fine details and degradation of contrast in the fused image. This deterioration in the image is associated with the conventional “averaging” rule for base layer fusion and relatively large feature extraction by CNN. To overcome these problems, an effective fusion framework based on visual saliency weight map (VSWM) combined with CNN is proposed. The proposed framework first employs VSWM method to improve the contrast of an image under consideration. Next, the fine details in the image are preserved by applying multi-resolution singular value decomposition (MSVD) before further processing by CNN. The promising experimental results show that the proposed method outperforms state-of-the-art methods by scoring the highest over different evaluation metrics such as Q0, multiscale structural similarity (MS_SSIM), and the sum of correlations of differences (SCD).

The False data injection attack (FDIA) against the Cyber-Physical Power System (CPPS) is a kind of data integrity attack. With more and more cyber vulnerabilities detected out, different types of FDIAs are emerging as severe threats to the stable operation of CPPS gradually. In this paper, the invasion pathway of the FDIA against CPPS is explored in detail, and a novel FDIA detection model based on ensemble learning is further provided. First, a pseudo-sample database is built to assist the training and evaluation of this model, and it's more important to update the model in the future. Furthermore, the optimal feature set is extracted to characterize the behavior of the FDIA, which improves the precision of the FDIA detection model. Finally, a focal-loss-lightgbm (FLGB) ensemble classifier is constructed to detect the FDIA behavior automatically and accurately. We illustrated the performance of this model by a fusion of measurement data and power system audit logs. This model utilizes the offline training way, the conclusion shows the high precision and stability of this model, which ensures the stable operation of the smart grid and improves the FDIA resistance ability of the CPPS.

Visibility restoration of an underwater image degraded by turbid water (due to scattering, absorption, and reflection) is a challenging task which requires careful design of computational imaging methods. The fundamental problems which limit image restoration are random diffusion and absorption of light in turbid water. Recently, the red-channel underwater image restoration method has emerged as an effective approach for visibility restoration under turbid water. However, this method fails to restore the visibility in a degraded image captured under non-uniform turbid water. The method fails due to its reliance on single transmission which cannot correctly express light propagation in an underwater environment. To overcome these problems, we propose a novel image restoration method based on adaptive color compensation and dual transmission estimation (ACDTE) to restore the visibility of underwater images degraded by non-uniform turbid water. The proposed method uses color-tone adaption to determine the hue of underwater images, and estimates global water light with quadtree decomposition. The method estimates dual transmission in the media through coefficient modification for R/G/B channels. Simulation and experimental results show that the proposed method can correct color deviations and has the advantage of visibility restoration of underwater images. The proposed method can be used for marine exploration, underwater rescue, and environmental monitoring.

Edge computing has been widely researched when 5G network and cloud platforms work together for people's life. The limitation of energy provided by the battery of the edge device hinders its application. This paper focuses on task scheduling in edge computing combined with the Energy harvesting technology (EH) and Energy Internet (EI). The edge node collects green energy by EH. And nodes exchange energy by EI. Energy Internet obtains green energy from edge nodes. Compared to green energy, we call energy from the power grid (not from the energy of edge nodes by energy harvesting technology) brown energy. How to reduce brown energy consumption is one of the most important problem in our paper. Previous works have not examined the energy attenuation between nodes, neither have they studied the immigration routes of virtual machines (VMs). This paper analyzes VM scheduling and models the energy consumption of VM immigrations, offloading tasks, and green energy transfer in edge computing. The paper proposes a heuristic assumption that there is only one VM in the system, and then presents three heuristics for the system with multiple VMs. The simulation results show that the proposed - immigrated VMs with the minimum energy transferring attenuation ratio method (METAR) is effective in reducing brown energy and total energy consumption, and improving the utilization rate of green energy. Compared to the Energy-Efficiency problem solution (EE-PRO) and maximize task energy consumption scheduling (MTS), METAR average reduces by 28.23% and 49.50% in brown energy consumption. At the same time, METAR average decreases by 5.67% and 11.52% in execution time.

To mitigate the conflict between imaging quality and speed, a spatially adaptive retina-like sampling method for 3-D imaging Lidar based on time-of-flight method is proposed. The differences between previous retina-like sampling method and the proposed method are described. Sampling points with dense distribution is for the area of interest while sparse distribution is for the area of uninterest, which obtains high imaging quality while consuming much less data acquisition time. Mathematical models of the spatially adaptive retina-like method are developed, and the key parameters are analyzed. To validate the spatially adaptive retina-like sampling method, we perform situational simulations to compare the proposed method with the previous one. Results demonstrate that the proposed method is capable of decreasing data acquisition time without considerable distortion of the interested target. Furthermore, the proposed method is analyzed under different scenes for single and multiple targets. Results illustrate that the proposed method performs better than the previous method.

A Fourier single-pixel imaging technique based on lateral inhibition is proposed to improve the reconstruction contrast and quality for low-contrast scenes. The mathematical models are developed, and corresponding simulations and experiments are conducted to verify and evaluate the features. Results show that this strategy strengthens the image contrast and structural similarity index mainly under low sampling ratios compared with the traditional Fourier single-pixel imaging. It offers an optional solution for various Fourier single-pixel imaging applications at low sampling ratios for low-contrast scenes.

Depth edge represents contours of depth images. However, captured images often have low resolution. Furthermore, conventional edge detection methods fail to handle depth images with low resolution. To cater depth images with low resolution, we propose a novel edge detection method with high accuracy and computational speed. The proposed method modifies the local binary pattern and selects an adaptive threshold to obtain a rough edge map. Then, it removes irrelevant edges to obtain the result. Experimental results show that the proposed method can rapidly acquire accurate depth edge maps and is applicable to depth images with low resolution. The proposed method is compared with and outperforms other renowned edge detection methods.

A retina-like 3-D (three-dimensional) computational ghost imaging (R-3DCGI) is proposed to obtain high resolution while maintaining high imaging efficiency. R-3DCGI models are developed, and corresponding simulations are conducted to verify and evaluate the accuracy and features of the R-3DCGI. Results show that the models obtain rotation and scaling invariances and compress redundant information, thereby improving imaging efficiency over the traditional computational ghost imaging. The relationships between parameters of the retina-like models and signal-noise ratio are discussed to optimize the retina-like structures in various applications.

Target tracking and imaging through scattering media are challenges that must be overcome for important applications in various fields. The fundamental problem is the randomly diffused light in scattering media that prevents the formation of diffraction-limited images. In recent years, speckle correlation has emerged as a powerful approach for restoring targets through scattering media and tracking targets with high scattering. However, this method fails to achieve high-accuracy target tracking in multidimensional motion. Furthermore, retrieving the deterministic phase of a target in a low-resolution speckle pattern is limited. Here, a method based on speckle-difference-combined bispectrum analysis (SDCBA) is presented to simultaneously track and image targets. The proposed SDCBA requires no high-resolution pattern, randomly assigned initial values, nor numerous iterations. The target is tracked simultaneously in multidimension by using speckle difference autocorrelation and restored with the deterministic phase via bispectrum analysis of the low-resolution speckle pattern. This work carries out simulations and experiments to demonstrate simultaneous multidimensional target tracking and imaging in low-resolution speckle pattern through scattering media via the SDCBA. This work will benefit various fields, including biomedical applications, materials science, and military security.

Image fusion integrates complex information about a target scene from multiple sensors into a single image. The fused image can further be utilized for human perception or different machine vision tasks. In the case of infrared and visible images, infrared images have the advantage of capturing thermal radiation intensity, whereas visible images are superior in gradient texture. In order to effectively fuse thermal intensity of infrared image and texture advantage of visible image, we propose a novel fusion method based on L0 decomposition and intensity mask. The proposed method first acquires base and detail layers of images (visible & infrared) using L0 decomposition. Next, an intensity mask is obtained using the basic global thresholding method on base layers of infrared image. The layers (base layers and detail layers) and visible images are divided images into three parts by the use of intensity mask, namely, mask-base layers, mask-detail layers, and texture-background. The first and second parts effectively achieve intensity blending, whereas the third part achieves the fused image with a clear gradient texture. The proposed method shows superior performance when compared with five state-of-the-art methods (on publicly available databases).

Softmax cross-entropy loss with L2 regularization is commonly adopted in the machine learning and neural network community. Considering that the traditional softmax cross-entropy loss simply focuses on fitting or classifying the training data accurately but does not explicitly encourage a large decision margin for classification, some loss functions are proposed to improve the generalization performance by solving the problem. However, these loss functions enhance the difficulty of model optimization. In addition, inspired by regularized logistic regression, where the regularized term is responsible for adjusting the width of decision margin, which can be seen as an approximation of support vector machine, we proposed a large-margin regularization method for softmax cross-entropy loss. The advantages of the proposed loss are twofold as follows: the first is the generalization performance improvement, and the second is easy optimization. The experimental results on three small-sample datasets show that our regularization method achieves good performance and outperforms the existing popular regularization methods of neural networks.

RF-based gesture sensing and recognition has increasingly attracted intense academic and industrial interest due to its various device-free applications in daily life, such as elder monitoring, mobile games. State-of-the-art approaches achieved accurate gesture sensing by using fine-grained RF signatures (such as CSI, Doppler effect) while could not achieve the same accuracy with coarse-grained RF signatures such as received signal strength (RSS). This paper presents rRuler, a novel feature extraction method which aims to get fine-grained human gesture features with coarse-grained RSS readings, which means rought ruler could measure fine things. In order to further verify the performance of rRuler, we further propose rRuler-HMM, a hidden Markov model (HMM) based human gesture sensing and prediction algorithm which utilizes the features extracted by rRuler as input. We implemented rRuler and rRuler-HMM using TI Sensortag platforms and off-the-shelf (CTOS) laptops in an indoor environment, extensively performance evaluations show that rRuler and rRuler-HMM stand out for their low cost and high practicability, and the average gesture sensing accuracy of rRuler-HMM can achieve 95.71% in NLoS scenario and 97.14% in LoS scenario, respectively, which is similar to the performance that fine-grained RF signatures based approaches could achieve.

Community detection is a classic and essential task in complex network analysis which aims at understanding their structural properties and functional organization. Community detection based on overlapping nodes has become one of the most popular methods in recent years. But, how to define the overlapping nodes in the real network is an important job. In this paper, we use the speaker-listener label propagation algorithm to find overlapping nodes. Second, we proposed a new metric which is based on local and global attributes to measure the node influence so as to evaluate the overlapping nodes’ importance. Community detection is realized with the $k$ -lowest-influence overlapping nodes deleting. Then, we assign the removed nodes into specific communities by voting to find the final community structure. The voting strategy is based on choosing the specific communities containing most of their neighbors. Finally, the extensive experiments on real-world networks demonstrate that our proposed method improves the quality of community detection methods and show both the effectiveness and efficiency of the method.

With the fast development of social networks, more and more data has been generated. Finding useful information among these data is important. Subgraph matching is the method that can be used in social networks for social search, recommender systems, and so on. In this paper, we choose five typical graph matching methods which are VF2, SPath, TurboISO, BoostIso, and RI to value their performance and scalability in social networks. These methods are verified by three social network data sets whose node’s number vary from thousands to millions. According to the experiments, we can find that VF2 and RI is applicable for rather small graphs while SPath performs better in large graph when average degree of graph is small. When graph with high average degree, TurboISO and BoostIso performs better. What is more, we also compare the efficiency of different matching order when searching. The order choosing strategy proposed by RI is better.

Nowadays, online product reviews play a crucial role in the purchase decision of consumers. A high proportion of positive reviews will bring substantial sales growth, while negative reviews will cause sales loss. Driven by the immense financial profits, many spammers try to promote their products or demote their competitors' products by posting fake and biased online reviews. By registering a number of accounts or releasing tasks in crowdsourcing platforms, many individual spammers could be organized as spammer groups to manipulate the product reviews together and can be more damaging. Existing works on spammer group detection extract spammer group candidates from review data and identify the real spammer groups using unsupervised spamicity ranking methods. Actually, according to the previous research, labeling a small number of spammer groups is easier than one assumes, however, few methods try to make good use of these important labeled data. In this paper, we propose a partially supervised learning model (PSGD) to detect spammer groups. By labeling some spammer groups as positive instances, PSGD applies positive unlabeled learning (PU-Learning) to study a classifier as spammer group detector from positive instances (labeled spammer groups) and unlabeled instances (unlabeled groups). Specifically, we extract reliable negative set in terms of the positive instances and the distinctive features. By combining the positive instances, extracted negative instances and unlabeled instances, we convert the PU-Learning problem into the well-known semisupervised learning problem, and then use a Naive Bayesian model and an EM algorithm to train a classifier for spammer group detection. Experiments on real-life Amazon.cn data set show that the proposed PSGD is effective and outperforms the state-of-the-art spammer group detection methods.