Search

A Scene-Based Nonuniformity Correction (SBNUC) approach is proposed for the correction of fixed-pattern-noise (FPN) in infrared image arrays. This method diverges from conventional neural-network based techniques by eschewing spatial filters for the computation of the desired image. Instead, it harnesses the sparse representation of the estimated irradiance within a hypersphere of adaptive radius, adhering to a distance constraint. The proposed method conceptualizes images as points in a high-dimensional space, with the distance between two images defined by their root mean square difference. The adaptive radius is derived from a sophisticated selection criterion, inversely related to the distance between the noisy input image and the estimated irradiance. This approach ensures edge preservation without excessive blurring while discarding FPN as redundant in the representation coefficients. We propose an adaptive learning rate tailored to the desired image from sparse representation, which contrasts with traditional learning rates that rely on local spatial standard deviation. This new learning rate is determined by the pixel-wise distance between the input image and the desired image, integrating a new motion detection criterion to synchronize the update of correction coefficients across pixels. Our experimental results demonstrate that the method proposed in this paper achieves better removal of fixed-pattern-noise in both simulated and real infrared image sequences compared to the methods used for comparison.

Vision-based smart agriculture is an important way to improve the efficiency of agricultural production. Labeling images for deep learning in complex field photos is a difficult task. In this paper, an image augmentation method that can help reduce the workload of image labeling based on synthetic images is proposed. The synthetic images consisted of three parts: crop, weed, and soil. The crop and weeds were obtained automatically by Excess Green (ExG) and minimum error threshold segmentation. The data augmentation method was tested on image classification, object detection, and semantic segmentation tasks by Resnet, YOLOV5, and DeeplabV3. The accuracy of the classification model reached 0.99. The IoU of object detection and semantic segmentation were 0.98 and 0.96, respectively. The results showed that the method in this paper was acceptable despite slight overfitting. This method was proposed based on the characteristics of field images, and it was meant for reducing the workload of labeling images.

In this article, a novel infrared maritime small target detection method, called local dissimilarity measure with anti-interference based on global graph clustering (LDMGGC), is proposed. The Wasserstein distance is introduced to calculate the dissimilarity of gray level distribution between a central region and its neighborhoods. These dissimilarities construct the feature of a region. With this feature, detection for recalling all suspected targets is achieved. As the maritime interferences among suspected targets are able to be clustered, relaxing mutual k nearest neighbor graph is introduced in global graph clustering for filtering interferences. With this method, real targets are detected and maritime interferences are filtered out. Experiments are conducted on three maritime datasets and a nonmaritime dataset for comparison. On three datasets, the proposed method achieves the best Receiver Operating Characteristic curves and Area Under Curve (0.99529, 0.99945, 0.99573, and 0.9906) values, meaning that the proposed method has high detection probability and low false-alarm ratio. Target Hit Rate (98.04%, 97.96%, 100%, and 99.24%) and Intersection of Union (0.8170, 0.7542, 0.5824, 0.7707) on four datasets of the proposed method show it has a strong ability to suppress the interferences.

Sea fog detection is a significant and challenging issue in meteorological satellite imagery. Distinguishing between sea fog and low clouds is challenging due to the similar morphology and brightness characteristics of these two phenomena on the imageries. Most of the existing deep learning methods are based on a single imagery feature extraction without the time-related features in imagery sequence. Although the designed temporal models, such as temporal U-Net, expand the available features from a single imagery to the consecutive frames and introduce general temporal information, the learned motion features are not explicit and can only be implicitly learned through a large amount of data. Thus, we introduce motion features obtained from continuous temporal imagery sequences into the sea fog detection task due to the discrepancy between sea fog and other types of clouds. In this article, under the motion features acquired by Horn–Schunck (HS) optical flow method and attention mechanisms, a Motion Attention Network (MoANet) for sea fog detection is proposed, named MoANet. We performed detailed experiments on the Himawaria-8 satellite imagery data set (H-8 Dataset). The Mean Intersection over Union (MIoU) of our method reaches 81.38%, which is 6.49% higher than the single imagery method. The visualization of the results shows that MoANet has more smooth edges, as well as detects more complete area than others. Furthermore, we validate on International Comprehensive Ocean-Atmosphere Data Set (ICOADS) through contrasting visibility value to prove the practicality of the proposed method and the accuracy achieves 90.65%.

This study aimed to improve the performance of single-channel steady-state visual evoked potential (SSVEP)-based visual acuity assessment by mode decomposition methods. Using the SSVEP dataset induced by the vertical sinusoidal gratings at six spatial frequency steps from 11 subjects, 3-40-Hz band-pass filtering and other four mode decomposition methods, i.e., empirical mode decomposition (EMD), ensemble empirical mode decomposition (EEMD), improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN), and variational mode decomposition (VMD), were used to preprocess the single-channel SSVEP signals from Oz electrode. After comparing the SSVEP signal characteristics corresponding to each mode decomposition method, the visual acuity threshold estimation criterion was used to obtain the final visual acuity results. The agreement between subjective Freiburg Visual Acuity and Contrast Test (FrACT) and SSVEP visual acuity for band-pass filtering (−0.095 logMAR), EMD (−0.112 logMAR), EEMD (−0.098 logMAR), ICEEMDAN (−0.093 logMAR), and VMD (−0.090 logMAR) was all pretty good, with an acceptable difference between FrACT and SSVEP acuity for band-pass filtering (0.129 logMAR), EMD (0.083 logMAR), EEMD (0.120 logMAR), ICEEMDAN (0.103 logMAR), and VMD (0.108 logMAR), finding that the visual acuity obtained by these four mode decompositions had a lower limit of agreement and a lower or close difference compared to the traditional band-pass filtering method. This study proved that the mode decomposition methods can enhance the performance of single-channel SSVEP-based visual acuity assessment, and also recommended ICEEEMDAN as the mode decomposition method for single-channel electroencephalography (EEG) signal denoising in the SSVEP visual acuity assessment.

A PCB spring terminal is a common type of connector found on the printed circuit board of electronic devices. Contact force is a crucial performance metric of this component that directly affects the product’s dependability. In actual production, the contact force is unstable, and the influencing factors are unknown, whereas conventional experimental research is restricted to costly and error-prone trial-and-error procedures. In this study, finite element simulation technology and Spearman rank correlation analysis are combined to determine the crucial factors controlling the contact force of the PCB spring terminals in the structural design parameters and the plating related parameters. The validity of the results is confirmed by the measured data. The results indicate that the thickness of the spring terminal is the most significant factor in controlling the contact force. According to the conclusion, the company has re-established the inspection standard and resolved production problems. This case shows that the combination of finite element simulation and correlation analysis can be used to find the most important design parameters of such components quickly, accurately, and cheaply.

Visible light communication (VLC) has been deemed to be one promising technique for the sixth-generation wireless communication network. However, due to the attenuation properties, the VLC ranges are usually limited. To unlock the advantage of multiple light-emitting diodes (LEDs) and extend the coverage of VLC systems, a relay-aided VLC system with non-orthogonal multiple access (NOMA) is proposed in this paper. Specifically, a decode-and-forward (DF) relay protocol is introduced to construct the relay-aided VLC system with NOMA. Moreover, the closed-form expression of the outage probability (OP) in the DF relay-aided VLC system with NOMA is derived to accurately characterize the theoretical performance of the system. Thereafter, the effects of the corresponding parameters on the OP are investigated. Simulation results verify the accuracy of the derived theoretical expression and illustrate that the DF relay-aided VLC system with NOMA can effectively improve system outage performance. Compared to the direct transmission NOMA VLC system without relay link, our scheme can obtain $\text{15}\,\text{dB}$ transmit signal-to-noise ratio gains in $10^{-3}$ OP level. Last but not least, the optimal transmit semi-angle and deployment height of the relay can be determined for achieving the optimal system outage performance.

Few-shot fine-grained image classification aims to solve the learning problem with few limited labeled examples. The existing methods use data augmentation to randomly transform the original examples to get new examples, and then use the new examples to train the model to improve the robustness and generalization ability of the learnt model. Due to each iteration of these methods uses a random transformation to get a new example, it will cause the unstable problem of the class center in the feature measurement stage. To solve this problem, a Multi-view Metric Learning (MML) method is proposed, which is based on a new concept (View Bag) and its effective similarity measurement method to achieve better few-shot fine-grained image classification. Firstly, a new example obtained by a kind of data augmentation is defined as a view, and a set of views generated by multiple data augmentation is defined as a view bag. Then, the view bag is sent into the model to extract the features, and a multi-view metric method with the view bag as the object is proposed to overcome the unstable problem of the class center. Finally, classification is performed by measuring the similarity between view bags. Experiments are conducted on three public datasets, CUB-2011-200, Stanford-Dogs and Stanford-Cars. The proposed method achieves 71.61±0.87%, 57.78±0.96% and 74.02±0.84% for the 5-way 1-shot classification task, and 88.72±0.51%, 76.30±0.68% and 92.94±0.37% for the 5-way 5-shot classification task, which have the state-of-the-art performances. Under the condition of the same backbone network, the proposed multi-view metric method can measure the similarity between examples more effectively, and improve the robustness and generalization ability of the model.

Urban rail transit has become an important traffic mode in large cities and has brought great conveniences to urban residents, however traffic accidents often occur in daily operations, therefore the safety and robustness must be paid more attention. In this paper, the network utilization of urban rail transit networks (URTNs) is expressed by the travel time difference between user equilibrium and optimal system, and the travel time difference is considered as the assessment indicator of robustness. Meanwhile, Shanghai metro network is taken as the example to analyze the robustness of URTNs under different disturbances, such as node attack, link-demand attack and overall-demand attack. The results show that URTNs have better robustness subjected to the random link-demand attack than node attack, and the robustness decreases with the increase of the attack intensity when the network suffers overall-demand attack. Moreover, we find that the critical stations and links can be identified, and URTNs can be paralyzed when the attack intensity exceeds the critical threshold.

The robust reconstruction of the reference signal is one of the key technologies in passive radars. This paper investigates the reference signal reconstruction problem for digital television terrestrial multimedia broadcasting (DTMB)-based passive radar. The conventional digital signal processing for the DTMB signal is usually based on the assumption of the equal bandwidth sampling (EBS) in which the sampling rate is equal to the nominal bandwidth 7.56 MHz defined in the DTMB standard. But we reveal that the high-frequency subcarriers of the DTMB signal are aliased under the EBS, which will degrade the reconstruction performance. Thus we suggest the oversampling approach for the DTMB-based passive radar. The least squares (LS) channel estimation under oversampling is thoroughly analyzed. Moreover to solve the problem that the LS algorithm cannot estimate the time-domain response of the channel under oversampling, we propose two modified LS channel estimation algorithms, namely the Tikhonov regularization LS and reduced rank approximation LS. The advantages of the reference signal reconstruction under oversampling over that under EBS are validated by both the simulation and field experimental data.

Although fine-grained image classification is able to classify more fine-grained sub-categories compared to its coarse-grained counterpart, it often fails to identify individual instances. Therefore, we propose a new instance-level image classification task which further refines the granularity of fine-grained classification in order to identify unique instances rather than a sub-category containing multiple instances. In addition, we introduce an instance-level image classification dataset, AircraftCarrier, which contains 20 global aircraft carrier classes, as the first publically available dataset for instance-level image classification. The classification of instance-level aircraft carriers can prove to be a challenging task due to large intra-category differences as well as variations in the camera view, illumination, scale, and the presence of complex backgrounds. The AircraftCarrier dataset put forward here has the potential to improve the development of instance-level image classification. At the same time, we provide a Simple Classification Head (SCH) technique for the classification of aircraft carriers, with classical convolutional neural network models as the backbone network. The SCH has better performance than a direct classification head, and these results provide a benchmark performance result for researchers. Furthermore, we evaluate several fine-grained image classification methods and give their benchmark results. Finally, we present the challenges of instance-level classification and discuss further directions. This study provides the first publicly available instance-level image classification dataset and a performance benchmark for further research. The dataset and codes can be downloaded at https://github.com/tsingqsu/AircraftCarrier_Dataset/.

Digital signage is an important outdoor advertising medium in cities. However, advertising on digital signage often lacks pertinence. Thus, it is important to introduce an accurate digital signage audience classification method to facilitate targeted advertising. In this study, a multi-label classification model based on a backpropagation (BP) neural network and the Huff model, referred to as the Huff-BP model, is proposed to investigate digital signage audience classification. A case study is performed on outdoor digital signage within the 6th Ring Road in Beijing, China, and economic census, population census, average housing price, social media check-in and the centrality of traffic networks as research data. The data are divided into 100 × 100-1,000 × 1,000 m normal grids. Multi-label classification modelling factors for various grid scales are constructed. The BP neural network classification algorithm is improved to solve the multilabel classification problem. In addition, an improved Huff model is used to calculate the digital signage influence values between each grid cell and integrated into the improved BP neural network to classify modelling factors at various scales. Finally, four metrics are used to examine the effectiveness of the proposed model. The results show that the Huff-BP-based multi-label classification model achieves relatively good classification results, and the digital signage audiences are mainly concentrated within the 4th Ring Road and near the 5th Ring Road.

Keyphrases provide core information for users to understand the document. Most previous works utilize machine learning based methods for keyphrases extraction and achieve promising performance. However, these methods focus on identify keyphrases from the input text, and can not extract keyphrases that do not appear in the text. In this paper, we present an encoder-decoder framework, which incorporating copying mechanism, to generate keyphrases for the given text. This framework (CopyNet) integrates the generation part and copying part. The generation part generates the keyphrase from the predefined vocabulary, and the copy part gets the keyphrases from the source text. Furthermore, we improve the CopyNet by using different probability of the two parts. To incorporate more related information for keyphrase generation, the automatically built keyphrase semantic web is merged into the dataset to participate in the training process of the neural network. Semantic similarity based and word co-occurrence based methods are used for keyphrase semantic web construction. We build a large-scale biomedical keyphrase dataset to evaluate the system performance. Experiments show that our improved CopyNet can achieve better performance with different portions of the generation and copying part, and the incorporation of the semantic web also effectively improves the keyphrase generation.

An interacting multiple model based on unscented Kalman filter (IMM-UKF) is widely applied to positioning and tracking targets in various tracking scenarios. At the same time, visible light positioning (VLP) is developing rapidly due to the low cost and accuracy. Therefore, indoor positioning and tracking based on VLP combined with the IMM-UKF has attracted considerable interest. However, existing algorithms work on the assumption that the light-emitting diodes used for tracking are all point light sources, which ignores the geometry of these transmitters and results in low tracking accuracy. To overcome this problem, this paper proposes an innovative tracking algorithm based on VLP. This algorithm considers the shapes of the lights in combination with existing tracking algorithms, such as IMM-UKF. Simulation results show that in a standard Gaussian noise environment, the larger the transmitter is, the more meaningful the proposed algorithm is.

Real-time road information plays a crucial role in enabling intelligent transportation systems (ITS) applications. With sufficient road information, the map of road topography can be built and updated more easily. Furthermore, many appealing ITS applications can be enabled accordingly. Aiming at improving the quality and update rate of road information, a hot topic today is how to mine information from global positioning systems (GPS) trajectories by the clustering-based methods. Such schemes, however, encounter two challenges: 1) GPS noise and 2) low sampling rate of GPS traces data. As a result, it is difficult to infer road information from these irregular clusters. To tackle the above issues, we directly mine useful road information, heading, and width of roads, for ITS applications from GPS point cloud, i.e., a set of GPS points. First, the distribution of GPS points is discussed and the least squares method (LSM) is demonstrated to be outstanding for mining the heading of the road under a huge number of GPS points. Second, the weighted approximation least squares method is proposed to improve the accuracy of the LSM. Furthermore, combining with relevant distribution features in GPS points, the data distribution variance-road width discrete model is proposed to mine road width from GPS point cloud. Finally, using real-world datasets, we demonstrate that these proposed methods can achieve satisfactory performance in practice.

The image caption is a technology that enables us to understand the contents and generate descriptive text, of images using machines. With the development of deep learning, means of using it to understand image content and generate descriptive text has become a hot research topic. This paper proposes a multilayer dense attention model for image caption. A faster recurrent convolutional neural networks (Faster R-CNN) is employed to extract image features as the coding layer, the long short-term memory (LSTM)-attend is used to decode the multilayer dense attention model, and the description text is generated. The model parameters are optimized using strategy gradient optimization in reinforcement learning. Use of dense attention mechanisms in the coding layer can effectively avoid the interference of non-salient information and selectively output the corresponding description text for the decoding process. The experimental results in the field of general images validate the model's good ability to understand images and generating text.

In order to ensure the information security of a network, network security devices have been widely deployed in the information and communication network. Based on the monitoring data collected by network security equipment, network security administrators analyze whether there is an attacking incident or abnormal situation in the current network environment. However, if the original monitoring data collected by network security equipment is tampered with by the attacker, then network security administrators will make a wrong judgment and delay the best time to take protective actions. Therefore, the reliability and authenticity of data generated by network security devices are of the utmost importance. In this paper, we first define a revised Euclidean distance formula for the D-S evidence theory and propose a dual test method for the authenticity of data based on algebra statistics and geometric trends, detecting any malicious tampering on these monitoring data. Utilizing a web application firewall (WAF) device, we carry out a series of experiments to verify the proposed method.

The application of fine-grained image classification can be problematic due to subtle differences between classes. The existing global feature-based methods have worse accuracies than regional feature-based methods, because regional feature-based methods focus on the determination of differentiated features within local regions. To learn more discriminative global features, in this paper, we proposed the use of L2 normalization to tackle a neglected conflict between the widely used metric loss (triplet loss) and classification loss (softmax loss) in global feature-based methods. Furthermore, a multi-level metric learning network (MMLN) is proposed for fine-grained image classification based on global features. In the MMLN, multi-level metric learning objectives and classification objectives are present at multiple high-level layers. The multi-level metric learning objectives work together to supervise the network in order to learn highly discriminative features. In addition, a new probability aggregation strategy (PAS) is proposed to produce a fused prediction by combining the multi-level predictive probabilities. Experiments were conducted on three standard fine-grained classification datasets (CUB-200-2011, Stanford Cars, and FGVC-Aircraft). Results demonstrated that our MMLN achieved accuracies of 88.0%, 94.6% and 92.4% respectively and outperformed state-of-the-art methods, substantially improving fine-grained classification tasks. Besides, gradient-weighted class activation mapping (Grad-CAM) shows that the MMLN is able to pay more attention to the discriminative local regions due to the application of multi-level metric learning.

Coronary artery disease (CAD) has been one of main causes of heart diseases globally. The electrocardiogram (ECG) is a widely used diagnostic tool to monitor patients' heart activities, and medical personnel need to judge whether there are abnormal heartbeats according to captured results. Therefore, it is significant to identify ECG signals accurately and fast. In this paper, a fast and accurate electrocardiogram (ECG) classification system based on deep learning is proposed. In our model, stacked denoising autoencoders (SDAE), as encoder, automatically learns semantic encoding of heartbeats without any complex feature extraction in unsupervised way. Then bidirectional LSTM (Bi-LSTM) classifier achieves classification of heartbeats with semantic encoding. SDAE implements noise-reduction while Bi-LSTM takes full advantage of temporal information in data. At the same time, this method relieves impacts from unbalanced data by employing cost-sensitive loss function. We validate our model on MIT-BIH Arrhythmias Database, SVDB and NSTDB respectively. Compared with state-of-art methods, the final result verify that this newly proposed method not only has high accuracy but also boosts classifying efficiency.

We experimentally report that angle-control dynamics in the nonlinear propagation of the images of the probe, generated four-wave mixing (FWM) and fluorescence signals beams in FWM process in a cascade three-level, as well as a two-level atomic systems. It is shown that, by changing the angles between pumping fields and dressing fields, respectively, the characteristics of the measured signals including the spectra intensity and spatial images can be controlled to display many effects, including the electromagnetically induced transparency and absorption (EIT and EIA, respectively) and Autler-Townes splitting. In particular, we demonstrate the angle-controlled switching from EIA to EIT in probe transmission or from enhancement to suppression in FWM signal. Additionally, we also illustrate the interaction between spatial characteristics and spectra intensities of weak fields. The studies can be very useful in producing all optical-signal-processing devices, such as spatial beam splitters, routers, and switches.