IET Signal Processing最新文献

As an essential part of intelligent transportation system (ITS), traffic forecasting has provided crucial role for traffic management and risk assessment. However, complex spatial–temporal dependencies, heterogeneity, dynamicity, and periodicity of traffic data influence the traffic forecasting performance. Consequently, we propose a novel effective gated spatial–temporal merged transformer (GSTMT) inspired by multimask and dual branch for accurate traffic forecasting in this paper. Specifically, we first conduct a concatenation of gated spatial static mask transformer (GSSMT) and gated spatial dynamic mask transformer (GSDMT) with residual network. The GSSMT and GSDMT evolve from the traditional transformer by making preferable modifications that include gated linear unit (GLU), multimask mechanism including static mask matrix (SMM) and dynamic mask matrix (DMM), and spatial attention (SA). Among them, GLU is to promote the performance of capturing spatial dependency, dynamicity, and heterogeneity due to advanced performance for controlling information flow through layers. Additionally, by developing multimask mechanism including two novel SMM and DMM, the proposed GSTMT can precisely model the static and dynamic spatial structure for effectively highlighting static dependency and dynamicity. And SA is injected for enhancing the ability of capturing spatial dependency of GSSMT and GSDMT. Secondly, we develop a dual-branch gated temporal transformer (DBGTT) for capturing temporal dependency, heterogeneity, dynamicity, and periodicity via incorporating the GLU and mixed time series decomposition (MTD) into traditional transformer. Similarly, we also introduce the GLU for empowering DBGTT with capability of capturing temporal dependency, dynamicity, and heterogeneity. In addition, MTD, which brings dual-branch mechanism, can enhance the DBGTT for capturing more detailed temporal information via exploiting global and periodic profile of traffic data. At last, some experiments, which are performed on several real-world traffic datasets, demonstrate the better results over classic traffic forecasting methods.

Instance segmentation is a task that involves pixel-level classification and segmentation of each object instance in images. Various CNN-based methods have achieved promising results in natural image instance segmentation. However, the noise interference, low resolution, and blurred edges bring more significant challenges for sonar image instance segmentation. To solve these problems, we propose the Effective Strategy for Sonar Images Instance Segmentation (ESSIIS). We introduce ASception, a new network combining Atrous Spatial Pyramid Pooling (ASPP) and Extreme Inception (Xception). By integrating this with ResNet and transforming traditional convolutions into deformable convolutions, we further improve the ability of the network to extract features from sonar images. Additionally, we incorporate a bidirectional feature fusion module to enhance information fusion. Finally, we evaluate the detection accuracy and segmentation accuracy of the proposed method on the public sonar image dataset and the self-constructed dataset. ESSIIS attains a detection accuracy of 0.981 and a segmentation accuracy of 0.951 on SCTD, further impressively achieving 0.986 in both metrics when appraised on our dataset. The evaluation results demonstrate that the proposed method is more accurate, robust, and considerable for sonar image detection and segmentation.

Reinforcement learning (RL) has been applied to financial portfolio management in recent years. Current studies mostly focus on profit accumulation without much consideration of risk. Some risk-return balanced studies extract features from price and volume data only, which is highly correlated and missing representation of risk features. To tackle these problems, we propose a weight control unit (WCU) to effectively manage the position of portfolio management in different market statuses. A loss penalty term is also designed in the reward function to prevent sharp drawdown during trading. Moreover, stock spatial interrelation representing the correlation between two different stocks is captured by a graph convolution network based on fundamental data. Temporal interrelation is also captured by a temporal convolutional network based on new factors designed with price and volume data. Both spatial and temporal interrelation work for better feature extraction from historical data and also make the model more interpretable. Finally, a deep deterministic policy gradient actor–critic RL is applied to explore optimal policy in portfolio management. We conduct our approach in a challenging non-short-selling market, and the experiment results show that our method outperforms the state-of-the-art methods in both profit and risk criteria. Specifically, with 6.72% improvement on an annualized rate of return, 7.72% decrease in maximum drawdown, and a better annualized Sharpe ratio of 0.112. Also, the loss penalty and WCU provide new aspects for future work in risk control.

To improve direction of arrival (DOA) estimation for coherently distributed sources under impulsive noise environments, a logistic-based adaptive factor is proposed to suppress the impulsive noise contained in the output signals of the array. The properties of this adaptive factor are derived. Furthermore, this adaptive factor is applied to subspace methods, and a novel DOA estimation algorithm is proposed. This novel algorithm ensures the boundedness of the signal and the noise subspaces while improving the DOA estimation accuracy and robustness. The experimental results demonstrate that the proposed algorithm outperforms existing algorithms in terms of resolution probability and estimation accuracy under impulsive noise environments.

Insulator faults are an important factor in causing outages and accidents in power transmission lines. In response to problems related to inefficient insulator positioning, limited robustness of insulator defect feature extraction methods, and the scarcity of defective insulator samples leading to poor classifier generalization, a method for insulator defect detection and recognition based on vision big-model transfer learning and a stochastic configuration network (SCN) is proposed. First, data augmentation methods, such as Mosaic and Mixup, are employed to mitigate overfitting in the YOLOv7 network. Second, StyleGanv3 adversarial generative networks are used to augment the dataset of defective insulators, which enhances dataset diversity. Third, a vision big-model transfer learning method based on DINOv2 is introduced to extract features from insulator images. Finally, an SCN classifier is used to determine the status of insulators. Experimental results demonstrate that the applied data augmentation methods effectively mitigate overfitting. YOLOv7 accurately detects insulator positions, and the use of the DINOv2 feature extraction method increases the accuracy of insulator defect recognition by 28.6%. Compared with machine learning classification methods, the SCN classifier achieves the highest accuracy improvement of 17.4%. The proposed method effectively detects insulator positions and recognizes insulator defects.

In this paper, a novel direction-of-arrival (DOA) estimation method is proposed for linear arrays on the basis of residual neural network (ResNet). The real parts, imaginary parts, and phase entries of the spatial covariance matrix from the on-grid angles are used as the input of ResNet for training, and the angular directions formulated as a multilabel classification task are predicted using the sample covariance matrix from the off-grid angles during the testing phase. ResNet demonstrates robustness in the scenarios on a fixed number of signals and a mixed number of signals. Simulation results show that ResNet can achieve significant performance in DOA estimation compared to multiple signal classification, estimation of signal parameters via rotation invariance techniques, convolutional neural network (CNN), and deep complex-valued CNN in low signal-to-noise ratio.

Manual acupuncture manipulation (MAM) is essential in traditional Chinese medicine treatment. MAM action recognition is important for junior acupuncturists’ training and education; however, there are obvious personalized differences in hand gestures among expert acupuncturists for the same type of MAM. In addition, during the MAM operations, the magnitude and frequency of the expert acupuncturists’ hand shape and relative needle-holding finger position changes are tiny and fast, resulting in difficulties in observing MAM action details. Thus, we propose a Spatial Multiscale Interactive Fusion MAM Recognition Network to solve the difficulties in MAM recognition. First, this paper presents an optical flow-based hand motion contour global feature extraction method for acupuncture hand shape. Second, to explore the motion rule between the needle-holding fingers during the MAM operations, we design a quantitative description method of the relative motion of the needle-holding fingers: an “interactive attention module,” which achieves feature fusion and mines the correlation between different scales of MAM action features. Finally, the proposed MAM recognition method was validated by 20 acupuncturists from the Beijing University of Traditional Chinese Medicine and 10 from the Beijing Zhongguancun Hospital who participated in the MAM video signal collection. The proposed recognition method achieves the highest average validation accuracy of 95.3% and the highest test accuracy of 96.0% for four typical MAMs, proving its feasibility and effectiveness.

The detection of small infrared targets is still a challenging task and efficient and accurate detection plays a key role in modern infrared search and tracking military applications. However, small infrared targets are difficult to detect due to their weak brightness, small size and lack of shape, structure, texture, and other information elements. In this paper, we propose a target detection method. First, to address the problem that the proximity of targets to high-brightness clutter leads to missed detection of candidate targets, a Gaussian differential filtering preprocessed image is used to suppress high-brightness clutter. Second, a density-peaked global search method is used to determine the location of candidate targets in the preprocessed image. We then use local contrast to the candidate target points to enhance the gradient features and suppress background clutter. The Facet model is used to compute multidirectional gradient features at each point. A new efficient surrounding symmetric region partitioning scheme is constructed to capture the gradient characteristics of targets of different sizes in eight directions, followed by weighting the candidate target gradient characteristics using the standard deviation of the symmetric region difference. Finally, an adaptive threshold segmentation method is used to extract small targets. Experimental results show that the method proposed in this paper has better detection accuracy and robustness compared with other detection methods.

In a cognitive satellite network (CSN) with GEO and LEO satellites, there is a large propagation losses between the sensing satellite and the ground station. The results of spectrum sensing from a single satellite may be inaccurate, which will create serious interference in the primary satellite system. Cooperative spectrum sensing (CSS) has become the key technology for solving the above problems in recent years. However, most of the current CSS techniques are model-driven. They are difficult to model and implement in CSNs since their detection performance is strongly dependent on an assumed statistical model. Thus, we propose a novel CSS scheme, which uses convolutional neural networks (CNNs), self-attention (SA) modules, long short-term memory networks (LSTMs), and soft fusion networks, called CSL-SFNet. This scheme combines the advantages of CNNs, SA modules, and LSTMs to extract the features of the input signals from the spatial and temporal domains. Additionally, the CSL-SFNet makes use of a novel soft fusion technique that improves detection performance while also considerably reducing communication overhead. The simulation results demonstrate that the proposed algorithm can achieve a detection probability of 90% when the signal-to-noise ratio is −20 dB; it has a shorter running time and always outperforms the other CSS algorithms.

In this paper, we present a double-intelligent reflecting surfaces (IRS)-assisted multiuser secure system where the inter-IRS channel is considered. In particular, we maximize the weighted sum secrecy rate of the system by jointly optimizing the beamforming vector for transmitted signal and artificial noise at the base station (BS) and the cooperative phase shifts of two IRSs, under the constraints of transmission power at the BS and the unit-modulus phase shift of IRSs. To tackle the nonconvexity of the optimization problem, we first convert the objective function to its concave lower bound by utilizing a novel successive convex approximation technique, then solve the transformed problem iteratively by applying an alternating optimization method. The Lagrange dual method, Karush–Kuhn–Tucker conditions, and alternating direction method of multipliers are applied to develop a low-complexity solution for each subproblem. Finally, simulation results are provided to verify the advantages of the cooperative double-IRS scheme in comparison with the benchmark schemes.