Search

Generalized Zero-shot Learning (GZSL) seeks to identify objects from both seen and unseen classes, relying solely on labeled samples from the seen classes. One of GZSL's main challenges is the problem of bias towards seen classes, i.e. , the model tends to classify all test samples as seen classes due to the data imbalance and over-fitting. To reduce the prediction bias, we proposed to train a novel M ultiple C riteria C alibration (MCC) model with C onfusion eli M ination contrastive E mbedding (CME). Specifically, we designed a novel divide-and-conquer calibration network by leveraging the learned multiple Class-Level and Instance-Level criteria to separate unseen samples from seen samples. To enhance feature separation, we further formulated supervised confusion-elimination contrastive constraints in the latent space to ensure the alignment properties of visual-semantic features and simultaneously encourage the learned features to distribute uniformly, providing a reasonable distribution for MCC and classification. After that, to adaptively mitigate bias, we can follow the learned criteria to calibrate the predictions of both the seen and unseen classes in the latent space, or leave the original predictions unchanged. Extensive experiments under GZSL settings show that the proposed method outperforms the state-of-the-art methods on four popular benchmarks. [ABSTRACT FROM AUTHOR]

Inventory is the basis of business activities; inventory management helps industries keep their inventories stocked with reasonable quantities, which ensures consumers demand while minimizing storage costs. The traditional manual inventory management has low efficiency and a high labor cost. In this paper, we used improved YOLOv3 to detect the cups stored on the warehouse shelves and counted their numbers to realize automated inventory management. The warehouse images are collected by the camera and transmitted to the industrial computer, which runs the YOLOv3 network. There are three feature maps in YOLOv3, the two smaller feature maps and the structure behind them are removed, and the k-means algorithm is used to optimize the default anchor size. Moreover, the detection range is limited to a specified area. Experiments show that, by eliminating those two feature maps, the network parameter is reduced from 235 MB to 212 MB, and detection FPS is improved from 48.15 to 54.88 while mAP is improved from 95.65% to 96.65% on our test dataset. The new anchors obtained by the k-means algorithm further improve the mAP to 96.82%. With those improvements, the average error rate of detection is reduced to 1.61%. Restricted detection areas eliminate irrelevant items to ensure the high accuracy of the detection result. The accurately counted number of cups and its change provide significant data for inventory management. [ABSTRACT FROM AUTHOR]

Recent advance on highly realistic AI‐synthesised video approaches makes it hard to distinguish whether a speaker in video is real. Many existing DeepFakes detection approaches can be invalidated by using them as supervision in adversarial training, which in turn improves the performance of DeepFakes. This makes the problem a dilemma. However, human can recognise the identity of familiar persons by observing their motion styles. Inspired by this, the paper proposes a novel method to recognise original speaker's identity clues in DeepFakes videos by learning individual motion styles. As a biological signature, motion styles can neither be used in the training process of DeepFakes nor be modified to deceive detection methods without reducing realistic performance. Also, this paper proposes a novel pipeline to continuously restore the original frames from DeepFakes videos without knowing the DeepFakes approach in advance. Based on above ideas, our scheme makes it possible to restore DeepFakes videos. Through training the cross‐modal transfer module, the appearance embedding can be inferred from the identity code. Extensive experiments are conducted to reveal the effectiveness of the method, it shows for the first time that the original persons can be identified automatically and the original video from DeepFakes videos can be generated. [ABSTRACT FROM AUTHOR]

Vector quantisation (VQ) is a widely used method for data compression or data clustering. This study proposes a fast codebook design method for image VQ. General particle swarm optimisation (PSO)‐K‐means hybrid methods for codebook design take too much time. To deal with this problem, the authors present a new clustering method, which takes a very short time and generates a pretty good codebook. This method contains four parts, i.e. pooling, PSO algorithm, reverse cast, and K‐means algorithm. Pooled images are used for the PSO process while original images are used for K‐means fine‐tuning, and these two processes connected by the reverse cast process. In the authors' experiment, this method can dramatically reduce the computation time by using big sizes of pooling window or enhance the codebook quality by using small sizes of pooling window. They can make the calculation time almost one‐tenth of that of the PSO‐K‐means hybrid method, and the scheme is also faster than the K‐means algorithm. Experimental results demonstrate that the main advantages of the proposed algorithm lie in the fact that it can reduce the computation time and enhance the quality of codebooks. [ABSTRACT FROM AUTHOR]

Many real-world complex systems consist of a set of basic units that are connected by different kinds of relationships. All types of such systems can be described by a multilayer network, where each link represents different types of interaction among the same set of nodes. In this paper, we present a general framework to characterize the influences (centrality) of layers. Furthermore, we propose two measures for layer centrality in terms of network connectivity under this framework. The basic idea of our measures consists in assigning more centrality value to layers that contribute more connectivity in a multilayer network. In other words, layers are more influential if more centrality values of links are assigned to them. We validate the measures on a real-world dataset of air transportation multilayer network and find that the measures are able to extract novel and useful information from the dataset. [ABSTRACT FROM AUTHOR]

Complex network response to node loss is a central question in different fields of science ranging from physics, sociology, biology to ecology. Previous studies considered binary networks where the weight of the links is not accounted for. However, in real-world networks the weights of connections can be widely different. Here, we analyzed the response of real-world road traffic complex network of Beijing, the most prosperous city in China. We produced nodes removal attack simulations using classic binary node features and we introduced weighted ranks for node importance. We measured the network functioning during nodes removal with three different parameters: the size of the largest connected cluster (LCC), the binary network efficiency (Bin EFF) and the weighted network efficiency (Weg EFF). We find that removing nodes according to weighted rank, i.e. considering the weight of the links as a number of taxi flows along the roads, produced in general the highest damage in the system. Our results show that: (i) in order to model Beijing road complex networks response to nodes (intersections) failure, it is necessary to consider the weight of the links; (ii) to discover the best attack strategy, it is important to use nodes rank accounting links weight. [ABSTRACT FROM AUTHOR]

How to predict the future popularity of a message or video on online social media (OSM) has long been an attractive problem for researchers. Although many difficulties are still ahead, recent studies suggest that temporal and topological features of early adopters generally play a very important role. However, with the increase of the adopters, the feature space will grow explosively. How to select the most effective features is still an open issue. In this work, we investigate several feature extraction methods over the Twitter platform and find that most predictive power concentrates on the second half of the propagation period, and that not only a model trained on one platform generalizes well to others as previous works observed, but also a model trained on one dataset performs well on predicting the popularity for other datasets with different number of observed early adopters. According to these findings, at least for the best features by far, the data used to extract features can be halved without loss of evident accuracy and we provide a way to roughly predict the growth trend of a social-media item in real-time. [ABSTRACT FROM AUTHOR]

Vector quantisation (VQ) shows a good performance for image coding with high‐compression ratios. However, there are many difficulties for image coding with VQ, especially the edge degradation and high‐computational complexity. To resolve these two problems, the authors propose a new coding method based on edge orientation patterns (EOPs) by classifying image blocks into nine classes according to their edge orientations. For colour image coding, 27 codebooks (nine for each colour component) are pre‐designed based on a series training images. In the encoding stage, an input colour image is decomposed into Y, Cb, and Cr components, and each component image is divided into non‐overlapping 4 × 4 blocks. For each block, eight edge orientation templates of size 4 × 4 are performed to determine its edge orientation. According to the edge orientation, each block is compressed by using the corresponding codebook. Essentially, the authors' scheme is a kind of classified VC (CVQ). Simulation results show that, their EOP‐based CVQ can largely improve the compression efficiency as well as speeding up the encoding process and it is sufficient to establish effectiveness of the authors' algorithm as compared with the existing techniques. [ABSTRACT FROM AUTHOR]

Recent advances in convolutional neural networks have achieved prominent results for object counting or locating in images. In these networks, some adopt detection‐based methods which require providing bounding‐box annotations for training and these methods usually cannot deal with the crowded cases well. Other networks adopt density regression‐based methods which require only point annotations for training but these methods can only predict the total count without providing any location information. In this Letter, the authors propose an attention guided convolutional neural network (AGNet) for both crowded object locating and counting using point annotation as supervision without annotated bounding boxes which are far more time‐consuming to label. Accordingly, they define a multi‐task objective loss function to lead the training of AGNet to achieve the goal of crowded object locating. The experimental results on four challenging benchmark datasets for crowded objects counting demonstrate the superiority of the proposed AGNet over recent state‐of‐the‐art approaches. [ABSTRACT FROM AUTHOR]

Controllability of complex networks has attracted much attention, and understanding the robustness of network controllability against potential attacks and failures is of practical significance. In this paper, we systematically investigate the attack vulnerability of network controllability for the canonical model networks as well as the real-world networks subject to attacks on nodes and edges. The attack strategies are selected based on degree and betweenness centralities calculated for either the initial network or the current network during the removal, among which random failure is as a comparison. It is found that the node-based strategies are often more harmful to the network controllability than the edge-based ones, and so are the recalculated strategies than their counterparts. The Barabási-Albert scale-free model, which has a highly biased structure, proves to be the most vulnerable of the tested model networks. In contrast, the Erdős-Rényi random model, which lacks structural bias, exhibits much better robustness to both node-based and edge-based attacks. We also survey the control robustness of 25 real-world networks, and the numerical results show that most real networks are control robust to random node failures, which has not been observed in the model networks. And the recalculated betweenness-based strategy is the most efficient way to harm the controllability of real-world networks. Besides, we find that the edge degree is not a good quantity to measure the importance of an edge in terms of network controllability. [ABSTRACT FROM AUTHOR]

Recently, as the controllability of complex networks attracts much attention, how to optimize networks’ controllability has become a common and urgent problem. In this paper, we develop an efficient genetic algorithm oriented optimization tool to optimize the controllability of arbitrary networks consisting of both state nodes and control nodes under Popov–Belevitch–Hautus rank condition. The experimental results on a number of benchmark networks show the effectiveness of this method and the evolution of network topology is captured. Furthermore, we explore how network structure affects its controllability and find that the sparser a network is, the more control nodes are needed to control it and the larger the differences between node degrees, the more control nodes are needed to achieve the full control. Our framework provides an alternative to controllability optimization and can be applied to arbitrary networks without any limitations. [ABSTRACT FROM AUTHOR]

Recently, the controllability of complex networks has attracted wide attention of researchers. The main contribution comes from Liu et al. who proposed the structural controllability as an analytical framework for making predictions regarding the control of directed networks in Nature. Since then, the controllability of many model and real networks has been deeply investigated except deterministic complex networks. In this paper, we focus on studying the controllability of deterministic complex networks. We examine six typical deterministic networks, the simulation results show that the minimum number of driver nodes grows linearly with network size. When the network size is large enough, the controllability approximates to a constant not more than 0.4, indicating that the deterministic networks are relatively easy to control. Furthermore, we investigate the characteristics of driver nodes in deterministic complex networks, finding that the driver nodes tend to avoid high degree nodes but to have high clustering coefficients. [ABSTRACT FROM AUTHOR]

This paper proposes an improved community model for social networks based on social mobility. The relationship between the group distribution and the community size is investigated in terms of communication rate and turnover rate. The degree distributions, clustering coefficients, average distances and diameters of networks are analyzed. Experimental results demonstrate that the proposed model possesses the small-world property and can reproduce social networks effectively and efficiently. [ABSTRACT FROM AUTHOR]

With the popularization of media-capture devices and the development of the Internet's basic facilities, video has become the most popular media information in recent years. The massive capacity of video imposes the demand of automatic video identification techniques which are very important to various applications such as content based video retrieval and copy detection. Therefore, as a challenging problem, video identification has drawn more and more attention in the past decade. The problem addressed here is to identify a given video clip in a given set of video sequences. In this paper, a robust video identification algorithm based on local non-negative matrix factorization (LNMF) is presented. First, some concepts about LNMF are described and the way of finding the factorized matrix is given. Then, its convergence is proven. In addition, a LNMF based shot detection method is proposed for constructing a video identification framework completely based on LNMF. Finally, a LNMF based identification approach using Hausdorff distance is introduced and a two-stage search process is proposed. Experimental results show the robustness of the proposed approach to many kinds of content-preserved distortions and its superiority to other algorithms. [ABSTRACT FROM AUTHOR]

In this paper, based on the phenomenon that individuals join into and jump from the organizations in the society, we propose a dynamic community model to construct social networks. Two parameters are adopted in our model, one is the communication rate Pa that denotes the connection strength in the organization and the other is the turnover rate Pb, that stands for the frequency of jumping among the organizations. Based on simulations, we analyze not only the degree distribution, the clustering coefficient, the average distance and the network diameter but also the group distribution which is closely related to their community structure. Moreover, we discover that the networks generated by the proposed model possess the small-world property and can well reproduce the networks of social contacts. [ABSTRACT FROM AUTHOR]

Video shot boundary detection (SBD) is the first and essential step for content-based video management and structural analysis. Great efforts have been paid to develop SBD algorithms for years. However, the high computational cost in the SBD becomes a block for further applications such as video indexing, browsing, retrieval, and representation. Motivated by the requirement of the real-time interactive applications, a unified fast SBD scheme is proposed in this paper. We adopted a candidate segment selection and singular value decomposition (SVD) to speed up the SBD. Initially, the positions of the shot boundaries and lengths of gradual transitions are predicted using adaptive thresholds and most non-boundary frames are discarded at the same time. Only the candidate segments that may contain the shot boundaries are preserved for further detection. Then, for all frames in each candidate segment, their color histograms in the hue-saturation-value) space are extracted, forming a frame-feature matrix. The SVD is then performed on the frame-feature matrices of all candidate segments to reduce the feature dimension. The refined feature vector of each frame in the candidate segments is obtained as a new metric for boundary detection. Finally, cut and gradual transitions are identified using our pattern matching method based on a new similarity measurement. Experiments on TRECVID 2001 test data and other video materials show that the proposed scheme can achieve a high detection speed and excellent accuracy compared with recent SBD schemes. [ABSTRACT FROM AUTHOR]

Abstract: In many real-life networks, both the scale-free distribution of degree and small-world behavior are important features. There are many random or deterministic models of networks to simulate these features separately. However, there are few models that combine the scale-free effect and small-world behavior, especially in terms of deterministic versions. What is more, all the existing deterministic algorithms running in the iterative mode generate networks with only several discrete numbers of nodes. This contradicts the purpose of creating a deterministic network model on which we can simulate some dynamical processes as widely as possible. According to these facts, this paper proposes a deterministic network generation algorithm, which can not only generate deterministic networks following a scale-free distribution of degree and small-world behavior, but also produce networks with arbitrary number of nodes. Our scheme is based on a complete binary tree, and each newly generated leaf node is further linked to its full brother and one of its direct ancestors. Analytical computation and simulation results show that the average degree of such a proposed network is less than 5, the average clustering coefficient is high (larger than 0.5, even for a network of size 2 million) and the average shortest path length increases much more slowly than logarithmic growth for the majority of small-world network models. [Copyright &y& Elsevier]

Abstract: As the deterministic version of the uniform recursive tree (URT), the deterministic uniform recursive tree (DURT) has been intensively studied by Zhang et al. (2008) . They gave several important properties of DURT, including its topological characteristics and spectral properties. Although DURT shows a logarithmic scaling with the size of the network, DURT is not a small-world network since its clustering coefficient is zero. In this paper, we propose a new deterministic small-world network by adding some edges with a simple rule in each DURT iteration, and then give the analytic solutions to several topological characteristics of the model proposed. [Copyright &y& Elsevier]

Abstract: In this paper, we propose a robust dual watermarking algorithm for video based on the audio video coding standard (AVS). One watermark is embedded in luminance components, whose embedding positions are optimized using the particle swarm optimization (PSO) technique, resisting common signal processing operations such as median filtering, adding noises, cropping and so on. To combat geometric and re-encoding attacks, the other watermark is embedded in chrominance components by adjusting the energy relationship between Cb and Cr components based on the just noticeable distortion (JND) concept of the human vision system (HVS). Both of them are implemented in the compressed domain by modifying the quantized coefficients, combined together to satisfy the requirements of robustness and imperceptibility. Only entropy decoding is required to detect the watermark, and thus the watermark detection process has low computational complexity. Extensive experimental results show that the high robustness and good quality of watermarked AVS video can be achieved simultaneously. [Copyright &y& Elsevier]

Proposes a method of embedding a gray-level digital watermark into an image. Methods for decomposing gray-level digital images; Experimental results; Conclusion.

At many construction sites, whether to wear a helmet is directly related to the safety of the workers. Therefore, the detection of helmet use has become a crucial monitoring tool for construction safety. However, most of the current helmet wearing detection algorithms are only dedicated to distinguishing pedestrians who wear helmets from those who do not. In order to further enrich the detection in construction scenes, this paper builds a dataset with six cases: not wearing a helmet, wearing a helmet, just wearing a hat, having a helmet, but not wearing it, wearing a helmet correctly, and wearing a helmet without wearing the chin strap. On this basis, this paper proposes a practical algorithm for detecting helmet wearing states based on the improved YOLOv5s algorithm. Firstly, according to the characteristics of the label of the dataset constructed by us, the K-means method is used to redesign the size of the prior box and match it to the corresponding feature layer to increase the accuracy of the feature extraction of the model; secondly, an additional layer is added to the algorithm to improve the ability of the model to recognize small targets; finally, the attention mechanism is introduced in the algorithm, and the CIOU_Loss function in the YOLOv5 method is replaced by the EIOU_Loss function. The experimental results indicate that the improved algorithm is more accurate than the original YOLOv5s algorithm. In addition, the finer classification also significantly enhances the detection performance of the model. [ABSTRACT FROM AUTHOR]

This Letter presents a new improved zero‐shot learning (ZSL) method based on meta‐learning that automatically adjusts metrics by multi‐layer perceptrons instead of using traditional Euclidean distance, cosine distance and other fixed metrics. In this way, the authors directly train the visual and semantic joint information of the images without adding additional auxiliary, which simply and effectively avoids the error derivation of unseen classes due to improper fixed metrics. Smooth one‐hot encoding rule is used to make the calculation of loss more reasonable. They also discuss the embedding methods to find a suitable low‐noise embedding space and analyse the model performance under the conventional setting and popular generalised ZSL setting. Experimental results show that the proposed automatic metrics method outperforms existing state‐of‐the‐art approaches in some parts of AwA2 and CUB datasets. [ABSTRACT FROM AUTHOR]

For covert communication in lossy channels, it is necessary to consider that the carrier of the hidden watermark will undergo multiple image-processing attacks. In order to ensure that secret information can be extracted without distortion from the watermarked images that have undergone attacks, in this paper, we design a novel fragmented secure communication system. The sender will fragment the secret data to be transmitted and redundantly hide it in a large number of multimodal carriers of messenger accounts on multiple social platforms. The receiver receives enough covert carriers, extracts each fragment, and concatenates the transmitted secret data. This article uses the image carrier as an example to fragment the text file intended for transmission and embeds it into a large number of images, with each fragment being redundant and embedded into multiple images. In this way, at the receiving end, only enough stego images need to be received to extract the information in each image, and then concatenate the final secret file. In order to resist various possible attacks during image transmission, we propose a strong robust image watermarking method. This method adopts a watermark layer based on DFT, which has high embedding and detection efficiency and good invisibility. Secondly, a watermark layer based on DCT is adopted, which can resist translation attacks, JPEG attacks, and other common attacks. Experiments have shown that our watermarking method is very fast; both the embedding time and the extraction time are less than 0.15 s for images not larger than 2000 × 2000 . Our watermarking method has very good invisibility with 41 dB PSNR on average. And our watermarking method is more robust than existing schemes and robust to nearly all kinds of attacks. Based on this strong robust image watermarking method, the scheme of fragmenting and hiding redundant transmission content into a large number of images is effective and practical. Our scheme can 100% restore the secret file completely under different RST or hybrid attacks, such as rotation by 1 degree and 5 degrees, scaling by 1.25 and 0.8, and cropping by 10% and 25%. Our scheme can successfully restore the secret file completely even if 30% of received images are lost. When 80% of received images are lost, our scheme can still restore 61.1% of the secret file. If all stego images can be obtained, the original text file can be completely restored. [ABSTRACT FROM AUTHOR]

A new feature for image retrieval based on edge orientation patterns (EOPs) by classifying image blocks into nine classes according to their EOs. Each input colour image is decomposed into Y, Cb and Cr components, and each component image is divided into non‐overlapping 4 × 4 blocks. For each block, eight EO templates of size 4 × 4 are performed to determine its EO. On the other hand, vector quantisation (VQ) is an efficient clustering and classification technique for high‐dimensional spaces, and thus VQ is also used to extract the texture features. The retrieval simulation results show that the EOP‐histogram‐based features combined with the traditional VQ‐index‐histogram‐based features can largely improve the recall and precision performance. [ABSTRACT FROM AUTHOR]

As one of the most fundamental tasks in computer vision, semantic segmentation assigns per pixel prediction of object categories. Training a robust model for semantic segmentation is challenging since pixel-level annotations are expensive to obtain. To alleviate the burden of annotations, the authors propose a weakly-supervised framework for zero-shot semantic segmentation, which can segment images having target classes without any pixel-level labelled instances. Under the assumption that the accessibility to image-level annotations of target classes does not violate the principle of zero pixel-level label in zero-shot semantic segmentation, we utilised image-level annotations to improve the proposed model's ability to extract pixel-level features. Furthermore, unlike existing zero-shot semantic segmentation methods, which use semantic embeddings as class embeddings to transfer knowledge from source classes to target classes, we use image-level features as their class embeddings to transfer knowledge since the distribution of pixel-level features is more similar to the distribution of image-level features rather than the distribution of semantic embeddings. Experimental results on the PASCAL-VOC data set under different data splits demonstrate that the proposed model achieves promising results. [ABSTRACT FROM AUTHOR]

Multi‐task cascaded convolutional neural network (MTCNN) is a human face detection architecture which uses a cascaded structure with three stages (P‐Net, R‐Net and O‐Net). The authors intend to reduce the computation time of the whole process of the MTCNN. They find that the non‐maximum suppression (NMS) processes after the P‐Net occupy over half of the computation time. Therefore, the authors propose a self‐fine‐tuning method which makes the control of computation time for the NMS process easier. Self‐fine‐tuning is a training trick which uses hard samples generated by P‐Net to retrain P‐Net. After self‐fine‐tuning, the distribution of human face probabilities generated by P‐Net is changed, and the tail of distribution becomes thinner. The control of the number of NMS input boxes can be made easier when the distribution has a thinner tail, and choosing a suitable threshold to filter the face boxes will generate less boxes. So the computation time can be reduced. In order to keep the performance of MTCNN, the authors still propose a landmark data set augmentation, which can enhance the performance of the self‐fine‐tuned MTCNN. From the experiments, it is found that the proposed scheme can significantly reduce the computation time of MTCNN. [ABSTRACT FROM AUTHOR]

Watermarking is an excellent solution to protect multimedia privacy but will be damaged by attacks such as noise adding, image filtering, compression, and especially scaling and cutting. In this paper, we propose a watermarking scheme to embed the watermark in the DWT-DCT composite transform coefficients, which is robust against normal image processing operations and geometric attacks. To make our scheme robust to scaling operations, a resampling detection network is trained to detect the scaling factor and then rescale the scaling-attacked image before watermark detection. To make our scheme robust to cutting operations, a template watermark is embedded in the Y channel to locate the cutting position. Experiments for various low- and high-resolution images reveal that our scheme has excellent performance in terms of imperceptibility and robustness. [ABSTRACT FROM AUTHOR]

In this study, the problem of event-triggered-based adaptive control (ETAC) for a class of discrete-time nonlinear systems with unknown parameters and nonlinear uncertainties is considered. Both neural network (NN) based and linear identifiers are used to approximate the unknown system dynamics. The feedback output signals are transmitted, and the parameters and the NN weights of the identifiers are tuned in an aperiodic manner at the event sample instants. A switching mechanism is provided to evaluate the approximate performance of each identifier and decide which estimated output is utilised for the event-triggered controller design, during any two events. The linear identifier with an auxiliary output and an improved adaptive law is introduced so that the nonlinear uncertainties are no longer assumed to be Lipschitz. The number of transmission times are significantly reduced by incorporating multiple model schemes into ETAC. The boundedness of both the parameters of identifiers and the system outputs is demonstrated though the Lyapunov approach. Simulation results demonstrate the effectiveness of the proposed method. [ABSTRACT FROM AUTHOR]

Portable document format (PDF) files are widely used in file transmission, exchange, and circulation because of their platform independence, small size, good browsing quality, and the ability to place hyperlinks. However, their security issues are also more thorny. It is common to distribute printed PDF files to different groups and individuals after printing. However, most PDF watermarking algorithms currently cannot resist print–scan attacks, making it difficult to apply them in leak tracing of both paper and scanned versions of PDF documents. To tackle this issue, we propose an invisible digital watermarking technology based on modifying the edge pixels of text strokes to hide information in PDFs, which achieves high robustness to print–scan attacks. Moreover, it cannot be detected by human perception systems. This method focuses on the representation of text by embedding watermarks by changing the features of the text to ensure that changes in these features can be reflected in the scanned PDF after printing. We first segment each text line into two sub-blocks, then select the row of pixels with the most black pixels, and flip the edge pixels closest to this row. This method requires the participation of original PDF documents in detection. The experimental results show that all peak signal-to-noise ratio (PSNR) values of our proposed method exceed 32 dB, which indicates satisfactory invisibility. Meanwhile, this method can extract the hidden information with 100% accuracy under the JPEG compression attack, and has high robustness against noise attacks and print–scan attacks. In the case of no attacks, the watermark can be recovered without any loss. In terms of practical applications, our method can be applied in the practical leak tracing of official paper documents after distribution. [ABSTRACT FROM AUTHOR]

In order to more effectively monitor and interfere with enemy signals, it is particularly important to accurately and efficiently identify the intercepted signals and estimate their parameters in the increasingly complex electromagnetic environment. Therefore, in non-cooperative situations, it is of great practical significance to study how to accurately detect direct sequence spread spectrum (DSSS) signals in real time and estimate their parameters. The traditional time-delay correlation algorithm encounters the challenges such as peak energy leakage and false peak interference. As an alternative, this paper introduces a Pseudo-Noise (PN) code period estimation method utilizing a one-dimensional (1D) convolutional neural network based on the residual network (CNN-ResNet). This method transforms the problem of spread spectrum code period estimation into a multi-classification problem of spread spectrum code length estimation. Firstly, the In-phase/Quadrature(I/Q) two-way of the received DSSS signals is directly input into the CNN-ResNet model, which will automatically learn the characteristics of the DSSS signal with different PN code lengths and then estimate the PN code length. Simulation experiments are conducted using a data set with DSSS signals ranging from −20 to 10 dB in terms of signal-to-noise ratios (SNRs). Upon training and verifying the model using BPSK modulation, it is then put to the test with QPSK-modulated signals, and the estimation performance was analyzed through metrics such as loss function, accuracy rate, recall rate, and confusion matrix. The results demonstrate that the 1D CNN-ResNet proposed in this paper is capable of effectively estimating the PN code period of the non-cooperative DSSS signal, exhibiting robust generalization abilities. [ABSTRACT FROM AUTHOR]

With the wide application of direct sequence spread spectrum (DSSS) signals, the comprehensive performance of DSSS communication systems has been continuously improved, making the electronic reconnaissance link in communication countermeasures more difficult. Electronic reconnaissance technology, as the fundamental means of modern electronic warfare, mainly includes signal detection, recognition, and parameter estimation. At present, research on DSSS detection algorithms is mostly based on the correlation characteristics of DSSS signals, and autocorrelation algorithm is the most mature and widely used method in practical engineering. With the continuous development of deep learning, deep-learning-based methods have gradually been introduced to replace traditional algorithms in the field of signal processing. This paper proposes a spread spectrum signal detection method based on convolutional neural network (CNN). Through experimental analysis, the detection performance of the CNN model proposed in this paper on DSSS signals in various situations has been compared and analyzed with traditional autocorrelation detection methods for different signal-to-noise ratios. The experiments verified the estimation performance of the model in this paper under different signal-to-noise ratios, different spreading code lengths, different spreading code types, and different modulation methods and compared it with the autocorrelation detection algorithm. It was found that the detection performance of the model in this paper was higher than that of the autocorrelation detection method, and the overall performance was improved by 4 dB. [ABSTRACT FROM AUTHOR]

While promising progress has been achieved in the zero-shot learning (ZSL) task. The existing approaches still suffer from the strong bias problem between the unseen and seen classes. This Letter presents a unified feature generating framework equipped with a boundary decision loss to tackle this issue in ZSL. Specifically, the hybrid semantic and visual classification strategy is proposed, which can effectively align the bidirectional visual-semantic interactions. Furthermore, this Letter introduces a decision loss that optimises the decision boundary of seen and unseen classes to further alleviate the confusion of generated features. Extensive experiments on three popular datasets animals with attributes, Caltech-UCSD-Birds 200-2011, and SUN show that the proposed approach outperforms previous state-of-the-art works under both traditional ZSL and challenging generalised ZSL settings. [ABSTRACT FROM AUTHOR]

A novel approach to detect and localise anomalous events in crowed scenes by processing surveillance videos is introduced in this study. Unusual events are those that significantly differ from current dominated behaviours. The proposed approach both detects pixel‐level and block‐level anomalies. In pixel level, Gaussian mixture models are used to detect abnormalities. Block‐level detection segments the crowd into blocks according to pedestrian detection, and then anomalies are spotted and localised with a social force model. Experimental results using the USCD datasets Ped1 and Ped2 show that the proposed method performs favourably against state‐of‐the‐art methods. [ABSTRACT FROM AUTHOR]

The purpose of image dehazing is to remove the interference from weather factors in degraded images and enhance the clarity and color saturation of images to maximize the restoration of useful features. Single image dehazing is one of the most important tasks in the field of image restoration. In recent years, due to the progress of deep learning, single image dehazing has made great progress. With the success of Transformer in advanced computer vision tasks, some research studies also began to apply Transformer to image dehazing tasks and obtained surprising results. However, both the deconvolution-neural-network-based dehazing algorithm and Transformer based dehazing algorithm magnify their advantages and disadvantages separately. Therefore, this paper proposes a novel Transformer–Convolution fusion dehazing network (TCFDN), which uses Transformer's global modeling ability and convolutional neural network's local modeling ability to improve the dehazing ability. In the Transformer–Convolution fusion dehazing network, the classic self-encoder structure is used. This paper proposes a Transformer–Convolution hybrid layer, which uses an adaptive fusion strategy to make full use of the Swin-Transformer and convolutional neural network to extract and reconstruct image features. On the basis of previous research, this layer further improves the ability of the network to remove haze. A series of contrast experiments and ablation experiments not only proved that the Transformer–Convolution fusion dehazing network proposed in this paper exceeded the more advanced dehazing algorithm, but also provided solid and powerful evidence for the basic theory on which it depends. [ABSTRACT FROM AUTHOR]

Abstract: Online video nowadays has become one of the top activities for users and has become easy to access. In the meantime, how to manage such huge amount of video data and retrieve them efficiently has become a big issue. In this article, we propose a novel method for video abstraction based on fast clustering of the regions of interest (ROIs). Firstly, the key-frames in each shot are extracted using the average histogram algorithm. Secondly, the saliency and edge maps are generated from each key-frame. According to these two maps, the key points for the visual attention model can be determined. Meanwhile, in order to expand the regions surrounding the key points, several thresholds are calculated from the corresponding key-frame. Thirdly, based on the key points and thresholds, several regions of interest are expanded and thus the main content in each frame is obtained. Finally, the fast clustering method is performed on the key frames by utilizing their ROIs. The performance and effectiveness of the proposed video abstraction algorithm is demonstrated by several experimental results. [Copyright &y& Elsevier]

With the increasingly extensive application of networking technology, security of network becomes significant than ever before. Encryption algorithm plays a key role in construction of a secure network system. However, the encryption algorithm implemented on resource‐constrained device is difficult to achieve ideal performance. The issue of power consumption becomes essential to performance of data encryption algorithm. Many methods are proposed to evaluate the power consumption of encryption algorithms yet the authors do not ensure which one is effective. In this study, they give a comprehensive review for the methods of power evaluation. They then design a series of experiments to evaluate the effectiveness of three main types of methods by implementing several traditional symmetric encryption algorithms on a workstation. The experimental results show that external measurement and software profiling are more accurate than that of uninterruptible power system battery. The improvement of power consumption is 27.44 and 33.53% which implies the method of external measurement and software profiling is more effective in power consumption evaluation. [ABSTRACT FROM AUTHOR]

Many centrality metrics have been proposed over the years to compute the centrality of nodes, which has been a key issue in complex network analysis. The most important node can be estimated through a variety of metrics, such as degree, closeness, eigenvector, betweenness, flow betweenness, cumulated nominations and subgraph. Simulated flow is a common method adopted by many centrality metrics, such as flow betweenness centrality, which assumes that the information spreads freely in the entire network. Generally speaking, the farther the information travels, the more times the information passes the geometric center. Thus, it is easy to determine which node is more likely to be the center of the geometry network. However, during information transmission, different nodes do not share the same vitality, and some nodes are more active than others. Therefore, the product of one node's degree and its clustering coefficient can be viewed as a good factor to show how active this node is. In this paper, a new centrality metric called vitality centrality is introduced, which is only based on this product and the simulated flow. Simulation experiments based on six test networks have been carried out to demonstrate the effectiveness of our new metric. [ABSTRACT FROM AUTHOR]

Abstract: This paper proposes a reversible data hiding method for natural images. Due to the similarity of neighbor pixels’ values, most differences between pairs of adjacent pixels are equal or close to zero. In this work, a histogram is constructed based on these difference statistics. In the data embedding stage, a multilevel histogram modification mechanism is employed. As more peak points are used for secret bits modulation, the hiding capacity is enhanced compared with those conventional methods based on one or two level histogram modification. Moreover, as the differences concentricity around zero is improved, the distortions on the host image introduced by secret content embedding is mitigated. In the data extraction and image recovery stage, the embedding level instead of the peak points and zero points is used. Accordingly the affiliated information is much smaller than in those methods of the kind. A sequential recovery strategy is exploited for each pixel is reconstructed with the aid of its previously recovered neighbor. Experimental results and comparisons with other methods demonstrate our method''s effectiveness and superior performance. [Copyright &y& Elsevier]

Abstract: Kernel discriminant analysis (KDA) is effective to extract nonlinear discriminative features of input samples using the kernel trick. However, the conventional KDA algorithm endures the kernel selection which has significant impact on the performances of KDA. In order to overcome this limitation, a novel nonlinear feature extraction method called adaptive quasiconformal kernel discriminant analysis (AQKDA) is proposed in this paper. AQKDA maps the data from the original input space to the high dimensional kernel space using a quasiconformal kernel. The adaptive parameters of the quasiconformal kernel are automatically calculated through optimizing an objective function designed for measuring the class separability of data in the feature space. Consequently, the nonlinear features extracted by AQKDA have the larger class separability compared with KDA. Experimental results on the two real-world datasets demonstrate the effectiveness of the proposed method. [Copyright &y& Elsevier]

Understanding the complex network response to nodes failure is important to protect them and guarantee their functioning. Most existing studies considered binary networks where the intensity (weight) of the links is not accounted for. However, in real world networks the intensity of the links widely differ so that a realistic description of complex networks is only possible if the heterogeneity present in edge weights is taken into account. Further, the vast majority of the existing studies investigates the efficacy of the attack strategies when all the nodes are removed. Despite it is important to study the network behavior in such a circumstance, real-world networks are more likely to experience local failures involving a small amount of nodes with a decrease in the system functioning not causing an overall collapse. This is the case of crossroad closing in complex road networks. For this reason it is interesting to investigate the networks response focusing to small node removals. In this paper we analyze the response of the real-world road networks of Beijing, the capital of China, to the removal of 1 and 2 nodes and accounting the weighted structure of the network. We use both classic binary node properties and network functioning measurement, introduced a weighted rank for node importance (node strength or page rank), and used a measurement for network functioning that accounts for the weight of the links (weighted efficiency). We find that the weighted node attack strategy Str (i.e. strength attack), which removes nodes according to their strength, is the most efficient to blemish the system when evaluating the damage with a weighted efficiency. Differently, betweenness attack strategy Bet , which removes nodes according to their centrality is the most efficient when using the binary efficiency to compute the network damage. These results indicate that adopting weighted or binary efficiency to measure networks functioning changed the efficacy of the attack strategies to damage the system. Last, we found a non-negligible fraction of random removals to be more efficient than the attack strategies; given that the main goal of the attack strategies is to individuate important nodes in the system, i.e. nodes which removal triggers highest damage, our results would indicate how in order to find the most important nodes may be more efficient to simulate an higher number of random nodes removal. [ABSTRACT FROM AUTHOR]

This study presents a hierarchical palmprint feature extraction and recognition approach based on multi‐wavelet and complex network (CN) since they can effectively decrease redundant information and enhance key points of main lines and wrinkles. The palmprint is first pre‐filtered and decomposed once using multi‐wavelet. Three components (LL1,2,3) corresponding to the pre‐filter except for diagonal component are extracted as the elementary features. Second, binary images (BLL1,2,3) are obtained by the average window method using different thresholds. Third, three series of dynamic evolution CN models (the 1st, 2nd, 3rd CNs) are constructed from global to local, which is based on the mosaiced images obtained from BLL1,2,3, BLL1 and four equally divided sub‐images of BLL1, respectively. Fourth, statistical features are extracted from complex networks, in which average degree and standard deviation of the degrees are extracted for the 1st CNs and average degrees are extracted for the 2nd and 3rd CNs. Fifth, the fisher feature is extracted using the linear discriminate analysis method. Finally, the nearest neighbourhood classifier is used to recognise palmprint. Based on the CASIA Palmprint Image Database, experimental results show that the proposed method can effectively recognise palmprint with good robustness and overcome the problem of small training samples number. [ABSTRACT FROM AUTHOR]

• We combined 1D and 2D convolutional kernels to extract orientation-aware features. • We employed a channel attention mechanism to adaptively select informative orientation-aware features. • Progressive feature fusion scheme is proposed to fuse hierarchical features. Single image super-resolution (SISR) is an active research topic in the fields of image processing, computer vision and pattern recognition, restoring high-frequency details and textures based on the low-resolution input image. In this paper, we aim to build more accurate and faster SISR models via developing better-performing feature extraction and fusion techniques. Firstly, we proposed a novel Orientation-Aware feature extraction/selection Module (OAM), which contains a mixture of 1D and 2D convolutional kernels (i.e., 3 × 1 , 1 × 3 , and 3 × 3) for extracting orientation-aware features. The channel attention mechanism is deployed within each OAM, performing scene-specific selection of informative outputs of the orientation-dependent kernels (e.g., horizontal, vertical, and diagonal). Secondly, we present an effective fusion architecture to progressively integrate multi-scale features extracted in different convolutional stages. Instead of directly combining low-level and high-level features, similar outputs of adjacent feature extraction modules are grouped and further compressed to generate a more concise representation of a specific convolutional stage for high-accuracy SISR task. Based on the above two important improvements, we present a compact but effective CNN-based model for high-quality SISR via Progressive Fusion of Orientation-Aware features (SISR-PF-OA). Extensive experimental results verify the superiority of the proposed SISR-PF-OA model, performing favorably against the state-of-the-art models in terms of both restoration accuracy and computational efficiency (e.g., SISR-PF-OA outperforms RCAN model, achieving higher PSNR 31.25 dB vs. 31.21 dB and using fewer FLOPs 764.41 G vs. 1020.28 G on the Manga109 dataset for scale factor × 4 SISR task.). The source codes will be made publicly available. [ABSTRACT FROM AUTHOR]

Complex networks are often subjected to failure and attack. Recent work has addressed the resilience of complex networks to either random or intentional deletion of nodes or links. Here we simulate the breakdown of the small-world network and the scale-free network under node failure or attacks. We analyze and discuss the dynamic correlation between degree and betweenness in the process of attack. The simulation results show that the correlation for scale-free network obeys a power law distribution until the network collapses, while it represents irregularly for small-world network. [ABSTRACT FROM AUTHOR]

The problem of finding the best strategy to attack a network or immunize a population with a minimal number of nodes has attracted much current research interest. The assessment of node importance has been a fundamental issue in the research of complex networks. In this paper, we propose a new concept called mapping entropy (ME) to identify the importance of a node in the complex network. The concept is established according to the local information which considers the correlation among all neighbors of a node. We evaluate the efficiency of the centrality by static and dynamic attacks on standard network models and real-world networks. The simulation result shows that the new centrality is more efficient than traditional attack strategies, whether it is static or dynamic. [ABSTRACT FROM AUTHOR]

Scale-free networks in which the degree displays a power-law distribution can be classified into assortative, disassortative, and neutral networks according to their degree-degree correlation. The second-order centrality proposed in a distributed computation manner is quick-calculated and accurate to identify critical nodes. We explore the second-order centrality correlation (SOC) for each type of the scale-free networks. The SOC-SOC correlation in assortative network and neutral network behaves similarly to the degree-degree correlation, while it behaves an apparent difference in disassortative networks. Experiments show that the invulnerability of most of scale-free networks behaves similarly under the node removal ordering by SOC centrality and degree centrality, respectively. The netscience network and the Yeast network behave a little differently because they are native disconnecting networks. [ABSTRACT FROM AUTHOR]

The invulnerability of complex networks is an important issue in that the behavior of scale-free network differs from that of exponential network. According to the structural characteristics of the networks, we propose two new attack strategies named IDB (initial degree and betweenness) and RDB (recalculated degree and betweenness). The strategies are originated from ID (initial degree distribution) and RD (recalculated degree distribution) strategies in which attacks are based on initial structural information of a network. The probability of node removals depends on a new metric combining degree centrality and betweenness centrality. We evaluate the efficiency of the proposed strategies on one real-world network and three network models. Experimental results indicate that the proposed strategies are more efficient than the traditional ID and RD strategies. Specially, the WS small-world network behaves more sensitive to the proposed strategies. The attack efficiency of RDB strategy is improved by 20% to RD strategy, and IDB strategy is improved by 40% to ID strategy. [ABSTRACT FROM AUTHOR]