Search

Agricultural pests significantly impact crop yield and quality, threatening food security and causing economic losses. Therefore, the precise identification of pests is crucial for improving agricultural production. However, traditional pest classification methods struggle to capture the complex relationships among different parts of pest images and often lack strong generalization capabilities, resulting in poor performance. To address these issues, we propose an agricultural pest classification model based on a multi-scale linear attention mechanism and Focal Loss. This model employs a multi-scale linear attention module to capture local features at various scales, as well as the long-distance dependencies and global relationships among these local features. It utilizes an attention mechanism with linear time complexity to ensure computational efficiency. In addition, we use the Focal Loss function to alleviate the impact of sample imbalance in the dataset and explore the effects of various data augmentation techniques on the model’s generalization ability. Experimental results demonstrate that our model performs excellently across datasets of different scales.

This paper tackles the exact consensus issue in nonlinearly parameterized multi-agent systems with unknown control directions, employing an adaptive learning control approach. The nonlinearly parameterized function is effectively compensated using neural networks and Fourier cardinal expansions. A cooperative consensus protocol is meticulously designed to achieve exact consensus through the application of iterative learning laws. The convergence of the consensus algorithm is rigorously analyzed using a composite energy function. Additionally, the robustness of the consensus algorithm is examined by introducing input disturbances to the controlled subsystem and bounded inputs to the leader. The scalability of the consensus protocol is explored through the context of formation control problems. Finally, the effectiveness of the constructed learning protocol is illustrated by simulations.

At present, in arbitrary-oriented object detection, the angular periodicity problem of rotated bounding box described by angle causes an object to have different numerical representations, which leads to uncertainty of rotated bounding box regression. To eliminate the angular periodicity problem, in this article, we propose a novel and simple ellipse parameters representation method for arbitrary-oriented object, which hides the angle of the object in the focal vector of ellipse to avoid direct angle prediction. Moreover, the proposed representation method can enable the arbitrary-oriented object to have only one numerical representation, which is beneficial to alleviate the uncertainty of bounding box regression. In order to adapt the proposed ellipse parameters representation method, we adopt 2-D Gaussian distribution label assign for coarse samples selection, then the Kullback–Leibler divergence loss and SimOTA are used to refine the coarse samples to obtain the best positive samples. We extend the YOLOX with medium parameters as an oriented ship detector according to the proposed ellipse parameters representation method, and conduct the experiments on HRSC2016, RSDD-SAR, and RHRSID to demonstrate the effectiveness of the proposed method. The experimental results show that the proposed representation method achieves impressive results compared with state-of-the-art arbitrary-oriented object detection methods.

This paper aims to establish a driving style recognition method that is highly accurate, fast and generalizable, considering the lack of data types in driving style classification task and the low recognition accuracy of widely used unsupervised clustering algorithms and single convolutional neural network methods. First, we propose a method to collect the information on driver’s operation time sequence in view of the imperfect driving data, and then extract the driver’s style features through convolutional neural network. Then, for the collected temporal data, the Long Short Term Memory networks (LSTM) module is added to encode and transform the driving features, to achieve the driving style classification. The results show that the accuracy of driving style recognition reaches over 93%, while the speed is improved significantly.

The low rank and sparse representation (LRSR) technique has attracted increasing attention for hyperspectral anomaly detection (HAD). Although a large quantity of research based on LRSR for HAD is proposed, the detection performance is still limited, due to the unsatisfactory dictionary construction and insufficient consideration of global and local characteristics. To tackle the above-mentioned concern, a novel HAD method, termed dual collaborative constraints regularized low-rank and sparse representation via robust dictionaries construction, is proposed in this article. Concretely, a robust dictionary construction strategy, which thoroughly excavates the potential of the density estimation model and local outlier factor, is proposed to yield pure and representative dictionary atoms. To fully exploit the global and local characteristics of hyperspectral images, dual collaborative constraints corresponding to the background and anomaly components are imposed on the LRSR model. Notably, two weighted matrices are further exerted on the representation coefficients to improve the effect of collaborative constraints, considering the fact that the surrounding pixels similar to the testing pixel should be given a large weight, otherwise the weight is expected to be small. In this way, the background and anomaly components can be well modeled. Additionally, a nonlinear transformation operation, which combines the output of the density estimation model and local outlier factor with the detection result derived from the LRSR model, is developed to suppress the background. The experiments conducted on one simulated dataset and three real datasets demonstrate the superiority of the proposed method compared with the four typical methods and four state-of-the-art methods.

Sparse representation (SR)-based approaches and collaborative representation (CR)-based methods are proved to be effective to detect the anomalies in a hyperspectral image (HSI). Nevertheless, the existing methods for achieving hyperspectral anomaly detection (HAD) generally only consider one of them, failing to comprehensively exploit them to further promote the detection performance. To address the issue, a novel HAD method, which integrates both SR and CR, is proposed in this article. To be specific, an SR model, whose overcomplete dictionary is generated by means of the density-based clustering algorithm and superpixel segmentation method, is first constructed for each pixel in an HSI. Then, for each pixel in an HSI, the used atoms in SR model are sifted to form the background dictionary corresponding to the CR model. To fully exploit both SR and CR information, we further combine the residual features obtained from both SR and CR model by the nonlinear transformation function to generate the response map. Finally, to preserve contour information of the objects, a postprocessing operation with guided filter is imposed into the response map to acquire the detection result. Experiments conducted on simulated and real datasets demonstrate that the proposed SRCR outperforms the state-of-the-art methods.

In practical applications, limited independent and identically distributed training snapshots brings a serious challenge in space-time adaptive processing (STAP), especially in the nonhomogeneous environments. Motivated by the significant spatial smoothness and sparsity commonality of weight vectors among related STAP tasks, we propose a novel STAP algorithm based on multi-task learning. In the proposed algorithm, the weight vectors corresponding to neighboring range bins of interest are kept consistent, and all weight vectors are constrained to share a common feature. Then, an alternating direction method of multipliers (ADMM) is used to solve the proposed algorithm, and the convergence of the algorithm is guaranteed. In addition, in case that the feature matrix is unknown or we want to learn a better feature matrix so that the associations among STAP tasks can be enhanced, we also provide an extension of the proposed algorithm to jointly optimize the feature matrix and weight matrix. Simulation results demonstrate the effectiveness of the proposed strategies.

Phase unwrapping (PU) has always been a critical and challenging step in interferometric synthetic aperture radar (InSAR) data processing. Inspired by existing research, i.e., the PGNet, we propose a novel quality-guided 2-D InSAR PU method via deep learning, and regard PU as a two-stage process. In the first stage, the ambiguity gradient is estimated using the proposed global attention U-Net (GAUNet) architecture, which combines the classic U-Net structure and global attention mechanism. Then, in the second stage, the classical PU framework (e.g., the L1- or L2-norm) is applied as a post-processing operation to retrieve the absolute phase. Since class imbalance is a key factor affecting the estimation of ambiguity gradient, different strategies based on four commonly used quality maps are adopted to deal with the problem. The quality map is not only input as additional information for the guidance of the training process, but also participates in the construction of loss function. As a result, GAUNet can pay more attention to the nonzero ambiguity gradients. By using the number of residues as the evaluation metric, we can choose the optimum strategy for the restoration of the absolute phase. In addition to the simulated interferograms, the proposed method is tested both on a real topographic interferogram exhibiting rugged topography and phase aliasing and a differential interferogram measuring the deformation from MW 6.9 Hawaii earthquake, all yield state-of-art performance when comparing with the widely used traditional 2-D PU methods.

Following man-made rules in the traditional control method of autonomous driving causes limitations for intelligent vehicles under various traffic conditions that need to be overcome by incorporating machine learning-based method. The latter is inherently suitable for simple tasks of autonomous driving according to its limited characteristic under complex multi-lane traffic conditions. In this paper, a decision control method is proposed based on multi-task reinforcement learning to address the shortcomings of autonomous driving control under complex traffic conditions. Herein, the autonomous driving task is divided into several subtasks utilizing the proposed method to reduce the training time and improve traffic efficiency under complex multi-lane traffic condition. To ensure the efficiency and robustness of agent convergence to the optimal action space, an adaptive noise exploration method is designed for the subtasks with convex characteristics. Five-lane driving tasks scenarios embedded in Carla simulator have been conducted to verify the proposed method. The results of the simulation draw the conclusion that the proposed method increases the driving efficiency of intelligent vehicles under complex traffic conditions.

Sector antenna azimuth plays an important role in the operation and maintenance of mobile networks. It is adjusted frequently so as to guarantee the high quality of coverage and low interference among neighboring cells. As one of the key elements of base station almanac (BSA), the azimuth of all the base stations need to be precisely managed by the operator. However, currently it is generally acquired by on-site measurement and updated manually, which is neither timely nor cost-effective. In addition, it is not open for third parties who need the information for network analysis. In view of this, by transforming the problem of azimuth estimation into one of searching for the optimal cell boundaries under the constraint of site location, this paper proposes an azimuth estimation method based on an improved multi-class soft-margin support vector machine (SVM). The explicit expression of the objective function of the problem is deduced through Lagrange duality. A circular one-versus-one strategy is utilized in dealing with the multi-class classification problem. In addition, a so-called average confusion index is designed to evaluate the degree of boundary separability of a base station so as to prejudge whether it is worthwhile to estimate its azimuths. Experiments are undertaken to validate the proposed algorithm by utilizing the spatial-temporal wireless signal dataset crowdsensed from massive user terminals in the 4G network. The experiments show that the proposed algorithm gives a significantly higher accuracy compared to that of two existing methods, namely the Gauss and radial rasterization methods. The impacts of data volume, penalty term and the boundary separability on the azimuth estimation is also discussed. The new algorithm is a promising alternative to the conventional manual solution with a much higher efficiency and lower cost, thus improving the intelligence of mobile network operation.

A robust adaptive chattering-free sliding mode (ACFSM) control method for electronic throttle (ET) system is proposed in this paper. It is well known that nonlinearities in the throttle system including friction, return-spring, limp-home (LH) and gear backlash affect the control accuracy of the throttle valve. Compared with the traditional sliding mode control methods, the ACFSM control not only overcomes the influence of nonlinearities and parameter uncertainties in the ET system, but also eliminates chattering in nature such that excellent throttle tracking performance and robustness are maintained. The proposed ACFSM control method is superior to the traditional sliding mode control in the following two aspects: 1) The chattering-free sliding mode control method is able to attenuate the control chattering without weakening the output tracking performance and robustness. 2) The upper bounds of the uncertainty and disturbance are not required any more in control design. They are online estimated by the adaptive law in the sense of Lyapunov. Moreover, due to the difficulty in selecting proper control parameters, the genetic algorithm (GA) is introduced to optimate the control parameters for the ACFSM controller prior to practical implementation. The comparative experimental results are given to demonstrate the excellent control performance of the proposed ACFSM control.

Network Representation Learning (NRL) is an effective way to analyze large scale networks (graphs). In general, it maps network nodes, edges, subgraphs, etc. onto independent vectors in a low dimension space, thus facilitating network analysis tasks. As community structure is one of the most prominent mesoscopic structure properties of real networks, it is necessary to preserve community structure of networks during NRL. In this paper, the concept of k-step partial community structure is defined and two Partial Community structure Guided Network Embedding (PCGNE) methods, based on two popular NRL algorithms (DeepWalk and node2vec respectively), for node representation learning are proposed. The idea behind this is that it is easier and more cost-effective to find a higher quality 1-step partial community structure than a higher quality whole community structure for networks; the extracted partial community information is then used to guide random walks in DeepWalk or node2vec. As a result, the learned node representations could preserve community structure property of networks more effectively. The two proposed algorithms and six state-of-the-art NRL algorithms were examined through multi-label classification and (inner community) link prediction on eight synthesized networks: one where community structure property could be controlled, and one real world network. The results suggested that the two PCGNE methods could improve the performance of their own based algorithm significantly and were competitive for node representation learning. Especially, comparing against used baseline algorithms, PCGNE methods could capture overlapping community structure much better, and thus could achieve better performance for multi-label classification on networks that have more overlapping nodes and/or larger overlapping memberships.

The global market share of electric vehicles (EVs) is on the rise, resulting in a rapid increase in their charging demand in both spatial and temporal domains. A remedy to shift the extra charging loads at peak hours to off-peak hours, caused by charging EVs at public charging stations, is an online pricing strategy. This paper presents a novel combinatorial online pricing strategy that has been established upon a reward-based model to prevent network instability and power outages. In the proposed solution, the utility provides incentives to the charging stations for their contributions in the EVs charging load shifting. Then, a constraint optimization problem is developed to minimize the total charging demand of the EVs during peak hours. To control the EVs charging demands in supporting utility's stability and increasing the total revenue of the charging stations, treated as a multi-agent framework, an online reinforcement learning model is developed which is based on the combination of an adaptive heuristic critic and recursive least square algorithm. The effective performance of the proposed model is validated through extensive simulation studies such as qualitative, numerical, and robustness performance assessment tests. The simulation results indicate significant improvement in the robustness and effectiveness of the proposed solution in terms of utility's power saving and charging stations' profit.

This paper proposes a real-time model-based health monitoring method for a nonlinear mechatronic system with multiple faults in both parametric and nonparametric components. A nonlinear bond graph model incorporating the influence of Stribeck friction is established to capture the dynamic behavior of the monitored mechatronic system. Based on the model, fault diagnosis is carried out via the combinative fault signature matrix which is built from independent and dependent analytical redundancy relations to enhance the isolation ability of the monitored system under multiple-fault condition. After that, an adaptive genetic algorithm-particle filter (AGA-PF) is developed for fault parameter estimation and remaining useful life prediction. The AGA-PF can mitigate the sample impoverishment problem in generic particle filter through genetic operators with adaptive mutation probability according to the fitness of the particles. The effectiveness of the proposed method is verified through simulation and experiment investigations on the nonlinear mechatronic system.

Decisions are usually made based on not only the performances of alternatives but also are implied by how the performances satisfy the decision makers' aspiration levels. This paper presents a linguistic aspiration-based solution to qualitative decision-making (QDM) where the aspiration levels and performances can be expressed by complex linguistic expressions (CLEs) such as hesitant fuzzy linguistic term sets and linguistic terms with weakened hedges. The proposed approach can deal with complex problems, which involve multi-criteria, multi-groups of experts, and multi-granular linguistic information. Based on the conventional aspiration-based approaches, the value function is defined by the probability of a CLE achieving its linguistic aspiration level. The performance of the proposed QDM approach is then demonstrated by solving the problem regarding the provider evaluation and selection. The proposed approach extends the range of available natural linguistic expressions that can be considered and used by experts and emphasizes the role of linguistic aspiration levels in QDM.

In this paper, energy management strategy (EMS) model based on deep recurrent neural network (DRNN) is presented to learn optimal torque distribution for the single-axle parallel hybrid electric vehicle. The model has two distinguishing properties: 1) because the EMS is formulated as a time series prediction problem, taking historical data as input of the EMS model captures the input-and-output dynamic characteristics and enhances the prediction capability and 2) the EMS model based on end-to-end framework directly generates torque distribution results without extracting features of driving cycles and other artificial interference. The extensive simulations are conducted to demonstrate the accuracy and generalization capability of the EMS model in public platform TensorFlow. Comparing with other energy management strategies, our proposed model yields better performance in terms of fuel economy and accuracy. The simulation results show that our proposed EMS model provides a novel way to study the energy management strategy.

In the era of 5G and beyond, heterogeneous network orchestration has become a tremendous issue. The dilemma facing future systems is how to allocate integrated resources to satisfy multifarious services, which is an imperative but arduous task in forming a systematic mathematical model and quantifying the model with its multi-layer uncertainty characteristics. Aiming at the statistical representation and optimization in multimodal heterogenous networks for 5G and beyond, we propose a novel hybrid probability process (HPP) as a generalized surrogate model and a weighted degenerated upper confidence bound (WDUCB) criterion for Bayesian optimization (BO). We apply the proposed HPP-WDUCB combination to our developed simulation platform and configure several applications of the integration of space information network in next generation communication systems. And we compared the proposed method with other surrogate models and acquisition strategies from a range of perspectives. The experiment results yield significant applicability and excellent performance in multimodal system representation and optimization which provides an effective statistical modeling and orchestration references for network tuning.

As energy supply on Unmanned Aerial Vehicle (UAV) is limited, energy efficient data transmission in UAV networks would be of great concern to the researchers. Nevertheless, UAV networks exhibit strong dynamic nature compared to ordinary Mobile Ad hoc Network (MANET), it is not rare that some nodes will be disjoint from other nodes from time to time. Under such circumstance, ordinary hop-by-hop routing schemes can not be used anymore, for there is no route existed from source to destination. Some solutions use store-carry-forward (SCF) routing to facilitate end to end data transmission. However, when should we use SCF routing, when we should use hop-by-hop routing if we taking energy into consideration? Is hop-by-hop routing always better than SCF? If it is good or not when the hop-by-hop routing is combined with the SCF routing? All these questions need to be answered. In this paper, we answered above questions by proposing three types of minimum energy consumption models of hop-by-hop routing, and two SCF routing models. We compared all these models and considered four tactics of combination modes for the transmissions in a rotary-wing UAV network. The researches show that two tactics have better performance than other modes in energy saving capabilities by reducing 70% at most with the hop-by-hop routing, and the transmission time of combination modes are in same scale compared with hop-by-hop routing. The two tactics are appropriate for different UAV networks respectively, in which the source UAV can or cannot get the GPS information of relay UAVs in the route. The research conclusion can also be used for the transmissions in a fixed-wing UAV network by modifying the energy consumption models of fixed wing UAV, which is different from rotary wing UAVs.

The energy dispersion of traditional bistatic multiple-input multiple-output (MIMO) radar severely degrades the angle estimation performance. To overcome this limitation, we propose to design the transmit beamspace matrix and apply the design result to jointly estimate the target directions of departure (DOD) and directions of arrival (DOA). The essence of this method is to minimize the distance between a desired transmit steering vector and the actual one within a given angular sector. Moreover, we enforce two inequality constraints to achieve approximately uniform power distribution across the transmit array elements. In addition, we carefully devise the desired transmit steering vector to expand the virtual transmit aperture and further improve the DOD estimation performance. Furthermore, we use tensor-decomposition and a phase-searching method to estimate the target DODs and DOAs. Finally, the simulation results demonstrate the effectiveness of the proposed algorithm.

Artificial intelligence (AI) is the most popular technology for searching the natural essence of human beings' intelligence. AI is influencing the paradigm of the enterprise management and product optimization. The development of AI in enterprise management provide an opportunity for all enterprises to construct automated business processes, improve the customer experience, and expand the product differentiation. Nowadays, the changeable world increases the level of difficulty to make an appropriate AI strategy for a company because of the information uncertainty and complexity. Probabilistic dual hesitant fuzzy set (PDHFS), which is a very effective tool to handle uncertain information, contains the hesitant fuzzy information and the corresponding probabilistic information. Standing in front of the more and more complex evaluation/selection problems, decision makers (DMs) could express their preference information more flexibly using the probabilistic hesitant fuzzy information. In this paper, we focus on the strategy selection problem in AI and solving it by a proposed integrated AHP and VIKOR method under probabilistic dual hesitant fuzzy information. First, we construct the probabilistic dual hesitant fuzzy comparison matrix (PDHFCM) and propose a specific transformation function for using AHP method under probabilistic dual hesitant fuzzy information. For completing the AHP, we redefine a consistency measure and propose an appropriate information-improved approach to obtain the consistent comparison matrix and the corresponding weight values simultaneously. In addition, we study the properties of PDHFS deeply and propose a new comparison method and a novel distance measure for PDHFS to distinguish the different probabilistic hesitant fuzzy information effectively. Then, we propose an integrated VIKOR and AHP method and use the method to solve the AI strategy selection problem. Finally, the availability and effectiveness of the proposed method are illustrated by a case on AI strategy selection.

In this paper, a novel methodology is proposed for low-light image enhancement. The proposed algorithm contains three stages: image reconstruction, image enhancement and color restoration. Two-dimensional compact variational mode decomposition (2D-TV-VMD) is employed to covert the RGB image into gray map through decomposing it on multiple gray eigenfunctions. A binary artifact indicator function is used to identify and eliminate potential artifact pixels in an image, and then low-light image enhancement via illumination map estimation (LIME) is used to enhance the reconstructed gray-scale map. Finally, color restoration is performed in RGB-color space to recover the color information. Subjective evaluation and objective evaluation of the proposed method, including no-reference image quality metric of contrast-distorted images based on information maximization (NIQMC), is conducted on different low-light images. Objective and subjective experimental performance demonstrate the competitive performance of the proposed algorithm compared with other state-of-art methods.

Interference management and energy management are two important issues in ultra-dense heterogeneous cellular networks (HetNets). However, load balancing investigated for system capacity on interference coordination is not efficient for energy saving in HetNets. Meanwhile, the maximum energy efficiency configuration leads to the serious unfair problem for users associated with larger power node for the ultra-dense scenario. Thus, in this paper, we investigate energy consumption jointly together with interference coordination for ultra-dense HetNets, and formulate max-min energy-efficient enhanced intercell interference coordination configuration problem. Due to the non-smooth and mixed programming nature of this formulation, we propose a novel iterative and distributed algorithm to solve the problem by using fractional programming and Lagrangian dual theory. The simulation results verified the effectiveness of our proposed algorithm and fairness achieved for energy efficiency of users, it especially identified a new energy efficiency tradeoff between macro user and small cell user with interference coordination in ultradense HetNets.

In this paper, a novel binary phase shift keying demodulator based on 1-D convolutional neural network (1-D CNN) is proposed. The utilization of neural networks to detect the locations of phase shifts in the modulated data distinguishes the proposed scheme from other neural network demodulators, which decide the symbols corresponding to the sampled data in a symbol period. Meanwhile, coordinating with the symbol synchronization algorithm, the proposed structure is able to deal with the carrier frequency offsets and sampling frequency errors. Compared to the conventional demodulators, the proposed 1-D CNN demodulator presents better bit error rates performance.

Existing machine-learning-based vehicle detection algorithms for intelligent vehicles have an obvious disadvantage in that the detection effect decreases dramatically when the distribution of training samples and the scene target samples do not match. To address this issue, a scene-adaptive vehicle detection algorithm based on a composite deep structure is proposed in this paper. Inspired by the Bagging (Bootstrap aggregating) mechanism, multiple relatively independent source samples are first used to build multiple classifiers and then voting is used to generate target training samples with confidence scores. The automatic feature extraction ability of deep convolutional neural network is then used to perform source-target scene feature similarity calculations with a deep auto-encoder in order to design a composite deep-structure-based scene-adaptive classifier and its training method. Experiments on the KITTI data set and a data set captured by our group demonstrate that the proposed method performs better than existing machine-learning-based vehicle detection methods. In addition, compared with existing scene-adaptive object detection methods, our method improves the detection rate by an average of approximately 3%.

In this paper, a digital-to-time converter (DTC) based on the principle of quantified phase shift resolution (QPSR) is proposed and tested. The QPSR principle is realized by analyzing the phase relationship between two periodic signals, which distinguishes the proposed DTC from conventional DTCs. For performance evaluation, the proposed DTC is implemented on Xilinx Virtex-5 and Virtex-6 field programmable gate array (FPGA) chip, respectively. The obtained resolution of the DTC implemented on Virtex-5 FPGA is 27.1 ps, and the differential nonlinearity (DNL) and integral nonlinearity (INL) are -0.269~+0.247 least significant bit (LSB) and -0.387~+0.171 LSB. Meanwhile, the DTC implemented on Virtex-6 FPGA can achieve a resolution of 3.93 ps, a dynamic range of 43 s, a DNL of -0.327~+0.358 LSB, and an INL of -2.6093~+2.4754 LSB. Experimental results prove that the proposed DTC features high-accuracy, low-cost, and easy implementation.

This paper presents an enhanced efficiency 3-D convolution operator based on optimal field programmable gate array (FPGA) accelerator platform. The proposed system takes advantages of the intermediate data delay lines, implemented in an FPGA, to avoid loading repetition of the input feature maps. This 3-D convolution accelerator performs 268.07 giga operations per second at 100-MHz operation frequency, with 330-mW power consumption. We experimentally demonstrate the enhanced efficiency of the proposed convolution accelerator, in comparison with the conventional technologies. The proposed 3-D convolution accelerator may find interesting applications in neural networks and video processing.

Night time pedestrian detection is more and more important in advanced driver assistant systems (ADAS). Traditional pedestrian detection algorithms in far infrared (FIR) images lack accuracy and have long processing times. Focusing on this issue, in this paper, a visual saliency-based pedestrian detection algorithm is proposed. First, areas that contain suspected pedestrians are detected using a fusion saliency-based method. Then, the sub-image of the suspected pedestrian is used as an input to a histogram of local intensity difference feature and cross kernel-based support vector machine classifier to make a final determination. Experiments performed using a real FIR road image data set demonstrated that the proposed fusion saliency-based region of interest (ROI) detection method has the largest pedestrian inclusion rate and the smallest ROI proportion compared with three other methods. Besides, compared with existing state-of-the-art pedestrian detection algorithms, the proposed method demonstrates a much higher pedestrian detection rate with a comparably short processing time.

Magnetic field effects on tris-(8-hydroxyquinoline) aluminum-based organic light-emitting devices (OLEDs) by employing Fe3O4 as a magnetic dopant in the hole-transport layer (HTL) have been studied. The magnetic doped OLEDs exhibit efficient injection and transport of holes, and its performances are further enhanced after a magnetic field is applied. The enhancement of luminance and current efficiency of 20% and 24% has been obtained from the magnetic doped devices, while they are only 8% and 9%, respectively, for the nondoped devices under an applied magnetic field of 500 mT. Organic magnetoresistance induced by the magnetic doped HTL is the main origin of increased electroluminescence for the magnetic doped OLEDs.