Driver fatigue recognition is a highly challenging issue because of the complexity of road conditions, the dynamics of traffic flow, and the differences between drivers. This article proposes a biologically inspired long short-term memory (LSTM) model with neural plasticity (NP-LSTM) to improve the learning and memory ability of the traditional driver fatigue recognition method, thus improving effectiveness and robustness of monitoring and early-warning systems. First, the approximate entropy (ApEn) of the time series of drivers’ operation behaviors and vehicle status is investigated to explore the features of potential irregularity in fatigue-driving behaviors; then, inspired by the plastic learning mechanism of biological neurons, the intrinsic plasticity and synaptic plasticity are embedded into the LSTM neural network to realize the classified storage of complex road patterns, the dynamics of traffic flow, and the memory of drivers’ individual differences; finally, the dropout technology is introduced to further build a “sparse” neural network, which avoids the repeated training of an unchanged neural network under different conditions and enhances the adaptability and generalization of the whole monitoring and early-warning system. Experimental study on real roads is conducted to demonstrate the effectiveness of the proposed method. The results show that the average recognition accuracy is 88.73%, demonstrating a better recognition performance of the proposed method.

Prosthetic attack is a problem that must be prevented in the current finger vein recognition application. To solve this problem, a finger vein living detection system was established in this article. The system first captures short-term static finger vein video by uniform near-infrared lighting, segments the veins by Gabor filters with current removing, calculates the multi-Scale spatial-Temporal maps(MSTmap) from the selected vein blocks, and trains the MSTmaps in the proposed Light-ViT network for the liveness detection. The MS maps are used to extract the coarse feature and Light-ViT is used to refine the liveness feature and predict the liveness result. Light-ViT, featuring an enhanced L-ViT backbone as its core, is constructed by interleaving multiple MN blocks and L-ViT blocks. This architecture effectively balances the learning of local image features, controls network parameter complexity, and substantially improves the accuracy of liveness detection. The accuracy of the Light-ViT network is verified to be 99.63% on the self-made living / prosthetic finger vein video dataset. This proposed system can also be directly applied to the finger vein recognition terminal after the model lightweighting.

In response to the challenges of narrow bandwidth, susceptibility to interference, and poor stability in shortwave communication, this study investigates the application of MIMO (Multiple Input Multiple Output) technology to the shortwave communication system. To address symbol interference arising from multipath effects, a time reversal operation is introduced, involving temporal inversion and complex conjugation of the signal. At the receiving end, matched filtering decoding is employed, resulting in a reliable and efficient 2×2 shortwave MIMO channel model. Through simulation validation, it has been demonstrated that time-reversed space-time coding can effectively enhance the system's bit error rate performance.

Vein segmentation and projection correction constitute the core algorithms of an auxiliary venipuncture device, responding to accurate venous positioning to assist puncture and reduce the number of punctures and pain of patients. This paper proposes an improved U-Net for segmenting veins and a coaxial correction for image alignment in the self-built vein projection system. The proposed U-Net is embedded by Gabor convolution kernels in the shallow layers to enhance segmentation accuracy. Additionally, to mitigate the semantic information loss caused by channel reduction, the network model is lightweighted by means of replacing conventional convolutions with inverted residual blocks. During the visualization process, a method that combines coaxial correction and a homography matrix is proposed to address the non-planarity of the dorsal hand in this paper. First, we used a hot mirror to adjust the light paths of both the projector and the camera to be coaxial, and then aligned the projected image with the dorsal hand using a homography matrix. Using this approach, the device requires only a single calibration before use. With the implementation of the improved segmentation method, an accuracy rate of 95.12% is achieved by the dataset. The intersection-over-union ratio between the segmented and original images is reached at 90.07%. The entire segmentation process is completed in 0.09 s, and the largest distance error of vein projection onto the dorsal hand is 0.53 mm. The experiments show that the device has reached practical accuracy and has values of research and application.

Prosthetic attack is a problem that must be prevented in current finger vein recognition applications. To solve this problem, a finger vein liveness detection system was established in this study. The system begins by capturing short-term static finger vein videos using uniform near-infrared lighting. Subsequently, it employs Gabor filters without a direct-current (DC) component for vein area segmentation. The vein area is then divided into blocks to compute a multi-scale spatial-temporal map (MSTmap), which facilitates the extraction of coarse liveness features. Finally, these features are trained for refinement and used to predict liveness detection results with the proposed Light Vision Transformer (Light-ViT) model, which is equipped with an enhanced Light-ViT backbone, meticulously designed by interleaving multiple MN blocks and Light-ViT blocks, ensuring improved performance in the task. This architecture effectively balances the learning of local image features, controls network parameter complexity, and substantially improves the accuracy of liveness detection. The accuracy of the Light-ViT model was verified to be 99.63% on a self-made living/prosthetic finger vein video dataset. This proposed system can also be directly applied to the finger vein recognition terminal after the model is made lightweight.

This paper proposes a deep learning scheme to automatically carry out reading recognition in wheel mechanical water meter images. Aiming at these early water meters deployed in old residential compounds, this method based on deep neural networks employs a coarse-to-fine reading recognition strategy, firstly, by means of an improved U-Net to locate the reading area of the dial on a large scale, and then the single character segmentation is performed according to the structural features of the dial, and finally carry out reading recognition through the improved VGG16. Experimental result shows that the proposed scheme can reduce the information interference of non-interested regions, effectively extract and identify reading results, and the recognition accuracy of 95.6% is achieved on the dataset in this paper. This paper proposes a new solution for the current situation of manual meter reading, which is time-consuming and labor-intensive, errors occur frequently; and the transformation cost is high and difficult to implement. It provides technical support for automatic reading recognition of wheel mechanical water meters.

With limited retrieval of reserves and restricted capability in plant pathology, automation of processes becomes essential. All over the world, farmers are struggling to prevent various harm from bacteria or pathogens such as viruses, fungi, worms, protozoa, and insects. Deep learning is currently widely used across a wide range of applications, including desktop, web, and mobile. In this study, the authors attempt to implement the function of AlexNet modification architecture-based CNN on the Android platform to predict tomato diseases based on leaf image. A dataset with of 18,345 training data and 4,585 testing data was used to create the predictive model. The information is separated into ten labels for tomato leaf diseases, each with 64 × 64 RGB pixels. The best model using the Adam optimizer with a realizing rate of 0.0005, the number of epochs 75, batch size 128, and an uncompromising cross-entropy loss function, has a high model accuracy with an average of 98%, a strictness rate of 0.98, a recall value of 0.99, and an F1-count of 0.98 with a loss of 0.1331, so that the classification results are good and very precise.

In order to improve the convenience and safety of the auto-guard system, a dual-model biometric recognition system of finger vein texture on dorsal is applied. Finger vein recognition is a living recognition technology, which can effectively prevent spoof attacks. This paper uses Phytium Ft-2000/4 development board as the embedded development platform, and the debain10 operating system is mounted on it to simulate the on-board system. Combined with the self-designed image acquisition module, the core functions of near-infrared imaging, region of interest extraction, dorsal finger vein and finger dorsal texture extraction, matching and recognition are realized and it is applied to the business system of auto-guard to simulate the interaction and application among users, door lock and on-board system. Experiments show that the system is convenient, safe and reliable, and has strong application and popularization value.

Emergency scheduling of public resources on the cloud computing platform network can effectively improve the network emergency rescue capability of the cloud computing platform. To schedule the network common resources, it is necessary to generate the initial population through the Hamming distance constraint and improve the objective function as the fitness function to complete the emergency scheduling of the network common resources. The traditional method, from the perspective of public resource fairness and priority mapping, uses incremental optimization algorithm to realize emergency scheduling of public resources, neglecting the improvement process of the objective function, which leads to unsatisfactory scheduling effect. An emergency scheduling method of cloud computing platform network public resources based on genetic algorithm is proposed. With emergency public resource scheduling time cost and transportation cost minimizing target, initial population by Hamming distance constraints, emergency scheduling model, and the corresponding objective function improvement as the fitness function, the genetic algorithm to individual selection and crossover and mutation probability were optimized and complete the public emergency resources scheduling. Experimental results show that the proposed method can effectively improve the efficiency of emergency resource scheduling, and the reliability of emergency scheduling is better.

In the field of advanced driver assistance systems (ADAS), effective learning of driver fatigue characteristics representation is a major challenge due to uncertainties of both real roads and drivers. To tackle this problem, this paper proposes a novel model of learning interpretable representations for fatigue features, so as to improve the performance of the monitoring system on real roads through the learned features. First, the approximate entropy of the Steering Wheel Angle (SWA)sequence is used to crop adaptive lengths of the Recurrent Neural Network (RNN) input data. Then, it will learn statistical indexes to discover the random steering characteristics. The Long Short-Term Memory (LSTM) unit can memorize drivers’ long-term operating characteristics and instantaneous change patterns, and mine their fatigue characteristics. Finally, the information gain method is used to discover the strong correlation between potential characteristics and fatigue levels, thereby obtaining the best feature representation for driver fatigue. The proposed method makes use of the advantages of recurrent neural network learning to explicitly capture various potential characteristics of fatigue driving and interactions between non-linear characteristics in driving. Experimental results on real driving data validate the effectiveness of our proposed method.

Plantar measurement has important application value in medicine and shoemaking industry. The existing plantar measurement technology generally has the problems such as complex operation process, long measurement time and high equipment price. This paper designs a plantar information acquisition and measurement system with convenient operation and low cost. The system can collect color images and depth images through structured light technology at the same time. By processing color image and depth image, the system segments the plantar image twice, obtains the plantar contour and feature points, and then calculates the plantar parameters. The results show that the system can measure human plantar information quickly and accurately, and has strong practicability.

In recent years, with the rapid development of biometrics technology, vein recognition is slowly integrating into our lives. At present, there are many related applications of hand veins and finger veins. The palm veins are deep under the skin and interfere with palm prints, which increases the difficulty of obtaining them, resulting in relatively few applications. Based on the research of palm vein image acquisition, this paper designs a set of auxiliary acquisition equipment to complete the acquisition of vein images under a comfortable somatosensory. The device takes the Raspberry Pi as the core of the model, supplemented by accessories such as luminous light source, optical sensor, control chip and small display, which can complete the collection of vein images. And through the algorithm of restricted contrast histogram equalization, Gaussian denoising, gabor filtering and other algorithms optimized for palm veins in the Raspberry Pi, the palm vein lines are enhanced to improve the image quality. The model integrates multiple modules into one mold, greatly reduces the volume of the model, improves the speed of the overall collection process, and has good application value.

This paper proposes an assumption that filtering out the confusing “awake” data from fatigue driving detection model promotes the accuracy of detection of “drowsy” status under real driving situation. Instead of focus on both “drowsy” and “awake” driving status, we set our first priority to alarm “drowsy” and temporarily ignore the accuracy of “awake” status recognition. The Support Vector Machine as a good classifier is employed for data filtering, provides more efficient training data and removes the data that may confuse the detection model. The results prove our assumption by 72% accuracy on “drowsy” recognition, which is higher than 38% recognition performed by detection without SVM filtering. In addition, the size of training samples after filtering for conducting detection model is extremely smaller than no filtering.

Camera-based vehicle detection under low illumination is a great challenge in intelligent transportation. Loud noise and insufficient colour information of the images hinder recognition quality of an interest target by traditional processing methods. To give full play to the strengths of bio-vision in graphic information processing and target recognition, this paper simulates the information processing and partition functions of its "what pathway", constructs a vision information processing model, and designs a computing method for image processing and recognition. Experiments under low illumination with 1,440 samples for vehicle detection have reached an average recognition rate of 92.1%, which proves this proposed method is able to improve the accuracy of vehicle detection by the intelligent transportation system.

This paper presents a contrast enhancement method based on fuzzy technique to infrared finger vein image for identification recognition. In order to solve the low contrast, blurring, speckle noise, the fuzzy set theory proposed is used for the captured infrared image enhancement. After the gray enhancement, the vein pattern feature extract and recognition are employed. The finger vein recognition accuracy results of 2070 vein images show the enhance efficient of the conducted fuzzy algorithm.

The study of the robust fatigue feature learning method for the driver’s operational behavior is of great significance for improving the performance of the real-time detection system for driver’s fatigue state. Aiming at how to extract more abstract and deep features in the driver’s direction operation data in the robust feature learning, this article constructs a fuzzy recurrent neural network model, which includes input layer, fuzzy layer, hidden layer, and output layer. The steering-wheel direction sensing time series sends the time series to the input layer through a fixed time window. After the fuzzification process, it is sent to the hidden layer to share the weight of the hidden layer, realize the memorization of the fatigue feature, and improve the feature depth capability of the steering wheel angle time sequence. The experimental results show that the proposed model achieves an average recognition rate of 87.30% in the fatigue sample database of real vehicle conditions, which indicates that the model has strong robustness to different subjects under real driving conditions. The model proposed in this article has important theoretical and engineering significance for studying the prediction of fatigue driving under real driving conditions.

Energy harvesting capability makes it possible to maintain network sustainability for wireless sensor networks. Due to the dynamics and instability of energy, it is significant to efficiently utilize the harvested part to keep the network sustainability. In this work, a Sustainable Clustering Protocol (SCP) is devised to provide continuous and timely data collection for sensor networks with ambient energy harvesting, such as solar power. In SCP, cluster heads (CHs) are only responsible for cluster coordination within clusters. In addition, cluster relays (CRs) are selected to share the data aggregation and long-haul forwarding responsibilities of original CHs. Consequently, network infrastructure is stably maintained since only intra-cluster CRs instead of CHs are selected periodically, which in turn decreases control overhead. Simulation results show that sustainable clustering can efficiently balance the energy consumption of all nodes and improve the sustainability of network compared to traditional clustering protocols.

Due to high temperature, high pressure, high corrosion and many other factors, the hazardous chemical device is facing more severe security challenges than other industries. Now, the monitoring methods have been very mature, which play a basic monitoring role, not a predictive fault diagnosis. In this paper, the hazardous chemical device's status data will been collected from the existing industrial monitoring network, the real-time data will be preprocessed and then stored in a database, and the data will be imported to the real-time data into the ontology model, the data will be performed by big data processing and automatic reasoning. So that real-time status of hazardous chemical device and the warning of security risks predict are easily got at any time. The model is proposed to solving the problem of knowledge representation and reasoning of the hazardous chemical device based on ontology. The model is analyzed and implemented in Protégé software.

The change of lighting conditions and facial pose often affects the driver's face's video registration greatly, which affects the recognition accuracy of the driver's fatigue state. In this paper, the authors first analyze the reasons for the failure of the driver's face registration in the light conditions and the changes of facial gestures, and propose an adaptive AAM (Active Appearance Model) algorithm of adaptive illumination and attitude change. Then, the SURF (speeded up robust feature) feature extraction is performed on the registered driver's face video images, and finally the authors input the extracted SURF feature into the designed artificial neural network to realize the recognition of driver's fatigue state. The experimental results show that the improved AAM method can better adapt to the driver's face under the illumination and attitude changes, and the driver's facial image's SURF feature is more obvious. The average correct recognition rate of the driver's fatigue states is 92.43%.

finger vein identification, as an important part in biological feature identification, has been widely used in various fields. The finger vein image shows the vein structure captured under infrared ray. This paper firstly ado pts the principal component analysis (PCA) to extract low-dimension features of vein images; and then constructs a multi-layer neural network classifier based on BP Neural Network, to realize the identity recognition. Experiments show that the proposed method reaches a satisfactory accuracy of 81% in recognizing 600 images, showing a great value in application.