基于Retinex增强的多尺度分解的红外与可见光图像融合算法

针对传统红外与可见光图像融合算法在低照度条件下融合结果对比度差、目标边际模糊、背景细节信息丢失等问题,提出一种基于Retinex增强的多尺度分解的红外与可见光图像融合算法。首先,利用Retinex对弱可见光图像进行单尺度Retinex(SSR)算法信息增强处理;其次,利用交叉双边滤波对源图像进行多尺度分解,先后得到图像的基层图像信息和细节层图像信息,对基础层图像采用绝对值取大策略和引导滤波相结合的融合方法,对细节层图像采用构建权重图和显著图的融合方法;最后,将处理后的基础层图像和细节层图像加权得到融合图像。从主观分析来看,本文方法可以有效提取和融合源图像中的重要信息,得到融合质量高、视觉效果自然清晰的图像。从客观评价来看,多组图像融合结果表明,本文所提方法的AG、SF、CE和FMI指标均较好。

Pseudo channel:time embedding for motor imagery decoding

Motor imagery(MI)based electroencephalogram(EEG)represents a frontier in enabling direct neural control of external devices and advancing neural rehabilitation.This study introduces a novel time embedding technique,termed traveling-wave based time embedding,utilized as a pseudo channel to enhance the decoding accuracy of MI-EEG signals across various neural network architectures.Unlike traditional neural network methods that fail to account for the temporal dynamics in MI-EEG in individual difference,our approach captures time-related changes for different participants based on a priori knowledge.Through extensive experimentation with multiple participants,we demonstrate that this method not only improves classification accuracy but also exhibits greater adaptability to individual differences compared to position encoding used in Transformer architecture.Significantly,our results reveal that traveling-wave based time embedding crucially enhances decoding accuracy,particularly for participants typically considered“EEG-illiteracy”.As a novel direction in EEG research,the traveling-wave based time embedding not only offers fresh insights for neural network decoding strategies but also expands new avenues for research into attention mechanisms in neuroscience and a deeper understanding of EEG signals.

Design of nuclear radiation level gauge based on normalized LMS filtering correction

In the measurement of liquid level in industrial site environment,noise interference can affect the measurement accuracy.In order to improve the measurement accuracy of liquid level in the viscous state,a nuclear radiation level measurement system based on the least mean square(LMS)filtering correction method is designed.The system uses STM32F103 as the control core and adopts HART bus HT1200M chip for remote signal transmission and reception.The adaptive LMS algorithm can be used for more accurate filtering,calculating iterative weight vector,updating weighted coefficient,effectively removing system measurement noise and improving the measurement accuracy.The results show that the nuclear radiation level gauge based on normalized LMS can correct the measurement system accuracy in adaptive rules,improve the measurement accuracy to meet the requirements of industrial field environment for liquid level measurement and enhance the industrial automation control degree.

Optimization of maintenance strategy for high-speed railwaycatenary system based on multistate model

A multi-objective optimization model considering both reliability and maintenance cost is proposed to solve the contradiction between reliability and maintenance cost in high-speed railway catenary system maintenance activities.The non-dominated sorting genetic algorithm 2(NSGA2)is applied to multi-objective optimization,and the optimization result is a set of Pareto solutions.Firstly,multistate failure mode analysis is conducted for the main devices leading to the failure of catenary,and then the reliability and failure mode of the whole catenary system is analyzed.The mathematical relationship between system reliability and maintenance cost is derived considering the existing catenary preventive maintenance mode to improve the reliability of the system.Secondly,an improved NSGA2(INSGA2)is proposed,which strengths population diversity by improving selection operator,and introduces local search strategy to ensure that population distribution is more uniform.The comparison results of the two algorithms before and after improvement on the zero-ductility transition(ZDT)series functions show that the population diversity is better and the solution is more uniform using INSGA2.Finally,the INSGA2 is applied to multi-objective optimization of system reliability and maintenance cost in different maintenance periods.The decision-makers can choose the reasonable solutions as the maintenance plans in the optimization results by weighing the relationship between the system reliability and the maintenance cost.The selected maintenance plans can ensure the lowest maintenance cost while the system reliability is as high as possible.

Effects of selective laser melting process parameters on powder formability of Ti6Al4V

Taking Ti6Al4V titanium alloy powder as the research object,on the basis of single layer scanning and single channel scanning experiment,this paper studies the influence of selective laser melting(SLM)process parameters on Ti6Al4V alloy material formability,and block forming experiment is carried out.Through the design of orthogonal experiment,morphology observation of sample and density analysis,results show that the best block molding parameters of SLM technology in Ti6Al4V alloy powder are laser power of 400 W,lap rate of 1 and the scanning speed of 750 mm/min,density can up to 96.17%.

Projectile impact point prediction method based on GRNN

In order to forecast projectile impact points quickly and accurately,aprojectile impact point prediction method based on generalized regression neural network（GRNN）is presented.Firstly,the model of GRNN forecasting impact point is established;secondly,the particle swarm algorithm（PSD）is used to optimize the smooth factor in the prediction model and then the optimal GRNN impact point prediction model is obtained.Finally,the numerical simulation of this prediction model is carried out.Simulation results show that the maximum range error is no more than 40 m,and the lateral deviation error is less than0.2m.The average time of impact point prediction is 6.645 ms,which is 1 300.623 ms less than that of numerical integration method.Therefore,it is feasible and effective for the proposed method to forecast projectile impact points,and thus it can provide a theoretical reference for practical engineering applications.

Fault diagnosis method of hydraulic system based on fusion of neural network and D-S evidence theory

According to fault type diversity and fault information uncertainty problem of the hydraulic driven rocket launcher servo system（HDRLSS） , the fault diagnosis method based on the evidence theory and neural network ensemble is proposed. In order to overcome the shortcomings of the single neural network, two improved neural network models are set up at the com-mon nodes to simplify the network structure. The initial fault diagnosis is based on the iron spectrum data and the pressure, flow and temperature（PFT） characteristic parameters as the input vectors of the two improved neural network models, and the diagnosis result is taken as the basic probability distribution of the evidence theory. Then the objectivity of assignment is real-ized. The initial diagnosis results of two improved neural networks are fused by D-S evidence theory. The experimental results show that this method can avoid the misdiagnosis of neural network recognition and improve the accuracy of the fault diagnosis of HDRLSS.

Uncertainty analysis of ship model vertical center of gravity and transverse moment of inertia test

The usability of test results of ship model vertical center of gravity and transverse moment of inertia is generally depends on its uncertainty. Referring to the guidelines for uncertainty analysis in examination of liquid dynamic recommended by International Towing Tank Conference （ ITTC）, the results were analyzed, bias limits and precision limits were calculated and total uncertainty was estimated. The total uncertainty of six tests on ship model vertical center of gravity is is 0. 16% of the mean value, and the total uncertainty of six tests on ship model transverse moment of inertia is 5.66% of the mean value. The test results show that the total uncertainty of both the multiple tests and the single test is from the precision limits of ship model vertical center of gravity and transverse moment of inertia tests. Thus, the improved measurement system stability can enormously decrease the total uncertainty of multiple tests and the single test.

Distribution of Fe and adsorption of NH_3 in mordenite:a computational study

The distribution of Fe and the adsorption of NH3 in H-[Fe]MOR （mordenite） were investigated using dispersion corrected density functional theory （DFT-D2）.Based on the results,it can be found that the most favorable site for the distribution of Fe is T1O6,followed by T2O5,T4O2 and T3O1,and energy differences for Fe in different T sites are less than 0.09 eV,indicating that Fe atoms may distribute in all kinds of T sites in MOR.In addition,the adsorption energies for NH3 at each crystallographic position of H-[Fe]MOR were also determined.Finally,it can be concluded that the Br（o）nsted acid site at T2O5 is stronger than the other acid sites,and the adsorption of NH3 on Br（o）nsted acid sites is more stable than on Lewis acid sites.

Research on dynamic error compensation for an armored K type thermocouple

A new method was proposed, in which a high-power CO2 laser modulated by high frequency was used as the driv- ing source to heat up a surface-temperature sensor. The continual beam and the pulsed beam sent out by the same laser could be used in the same system to carry on the static calibration of the radiation thermometer and the dynamic calibration of the temperature sensor to be checked. The frequency-response characteristics of high-speed radiation thermometer surpassed that of the temperature sensor, therefore it could be used as the reference value to calibrate the latter and let system error be cor- rected. Differences in the environment of the sensor installing and the error caused by the change of thermo-physical proper- ty could be avoided. Thus, the difficult problem of traceable dynamic calibration of temperature was solved. In experiment, to obtain the frequency characteristics of the thermocouple and the dynamic performance of the K type thermocouple, which could compensate the dynamic characteristics of the sensor, the sensor was dynamically corrected by using the method, and then the mathematical model was established.

High-precision grating sensor demodulation device's network interface based on DM9000A

High-precision fiber Bragg grating se nsor demodulation instrument with wide-range dynamic scanning can effectivel y improve the measuring range of the optical fiber grating sensor.Ethernet com munication module is an extremely important part of the high-precision grating sensor demodulation device.Network interface based on Ethernet control chip D M9000A is used to send and receive the Bragg grating sensing pulse.The network transformer YL18-2050S is used to convert and filter the pulse from network.The transmitting and receiving program of grating demodulation,hardware circuit of Ethernet transmission interface are designed.The experimental results show that the network interface can achieve accurate and real-time transmissi on of the grating sensing information at high speed.

Design of High-voltage Electrostatic Generator Based on STM32

This paper intreduced a high-precision high-voltage elec- trostatic generator which utilized STM32F103 as the main coutroller. The hardware and software design of the system were detailed. The full use of ample on-chip resources of STM32F103, such as ADC and the PWM output of timer, contributed to the small size and low cost of the system. The 16-bit PWM signals, generated by the timer on chip, served to adjust the our-put voltage accurately. The tot~ screen was responsible for the setting and display of output voltage, and the friendly humawcomptaer interaction was built. Experimental results indicated that this high-voltage static generator was of high precision and great practicability for application.

New Intellectual Economized Technique on Electricity Based on DSP

In order to resolve the problean of the unbalanced threephase and unstable voltage, intellectual economized technique on elec- tricity based on electromagnetic regulation and control is proposed in this paper. We choose the TMS320LF2407A as the control chip and stepper motor as the executing agency. The equipment controls the movable contact reaching to the assigned position on the magnetic coil quickly and accurately, and outputs the sine-wave voltage steadily along with the network voltage variation though the fuzzy Porportional Integral Derivative（PID） control algorithm of integral separation and incremental mode with setting dead area. The principle of work and the key technique on the electromagnetic regulation and control are introduced in detail in this paper. The experiment result gives a proof for all the algorithm mentioned in this paper.

Interface Design Of Digital Platform For Bio Signal Processing

Bio-sensor arrays for multi-channel recording have been developed recently and signal processing platforms for those signals have been studied actively.But it’s thereal situation which these technologies are generally developed and studied respectively.So the interface design between recording array and signal processing platform is also an important issue to make bio-sensor signal processing system.In this paper,we proposed interface which has unique protocols to control sensor array and operate platform.There are two types of protocols in the interface.One is between sensor array and MCU in platform and the other is between MCU and board for wireless communication.Basically,each protocol has two kinds of modes(single,frame)and it can be extended if needed.

Simulation on dynamic characteristics of TC4 cutting with crack defects

Titanium alloys play an important role in aerospace and other fields.However,after precision forging and cold rolling process,some defects will appear on the subsurface of titanium alloy bars,thus reducing the surface quality and precision of turning process.This study aimed at exploring the effect of crack defects on TC4 cutting.Firstly,the finite element cutting simulation model of TC4 material with crack defects was established in ABAQUS.Then,the cutting parameters such as cutting force,stress concentration,chip morphology,residual stress were obtained by changing the variables such as the size and height of crack defects.Finally,the turning experiment was carried out on centerless lathe.The results show that the cutting force changes abruptly when the defect position is located on the cutting path,the maximal stress occurs at the tip of the defect,and the mutation of stress value is more serious with the increase of defect size;the buckling deformation of chip morphology occurs and becomes less serious with the increase of the distance between the defect position and the workpiece surface;the surface residual stress near the defect is related to the stress when the tool is close to the defect,the larger defect size and the closer to the machined surface,the greater the residual stress.Therefore,under certain processing conditions,the TC4 material should avoid large size defects or increase the distance between defects and the machined surface,so as to obtain better and stable surface quality.

Automatic road extraction framework based on codec network

Road extraction based on deep learning is one of hot spots of semantic segmentation in the past decade.In this work,we proposed a framework based on codec network for automatic road extraction from remote sensing images.Firstly,a pre-trained ResNet34 was migrated to U-Net and its encoding structure was replaced to deepen the number of network layers,which reduces the error rate of road segmentation and the loss of details.Secondly,dilated convolution was used to connect the encoder and the decoder of network to expand the receptive field and retain more low-dimensional information of the image.Afterwards,the channel attention mechanism was used to select the information of the feature image obtained by up-sampling of the encoder,the weights of target features were optimized to enhance the features of target region and suppress the features of background and noise regions,and thus the feature extraction effect of the remote sensing image with complex background was optimized.Finally,an adaptive sigmoid loss function was proposed,which optimizes the imbalance between the road and the background,and makes the model reach the optimal solution.Experimental results show that compared with several semantic segmentation networks,the proposed method can greatly reduce the error rate of road segmentation and effectively improve the accuracy of road extraction from remote sensing images.

Effects of temperature, particle size, and air humidity on sensibility of typical high-energetic explosives

The production and utilization of high-energetic explosives often pose a range of safety hazards,with sensitivity being a key factor in evaluating these risks.To investigate how temperature,particle size,and air humidity affect the responsiveness of commonly used high-energetic explosives,a series of BAM(Bundesanstalt für Materialforschung und-prüfung)impact and friction sensitivity tests were carried out to determine the critical impact energy and critical load pressure of four representative high-energetic explosives(RDX,HMX,PETN and CL-20)under different temperatures,particle sizes,and air humidity conditions.The experimental findings facilitated an examination of temperature and particle size affecting the sensitivity of high-energetic explosives,along with an assessment of the influence of air humidity on sensitivity testing.The results clearly indicate that high-energetic explosives display a substantial decline in critical reaction energy when subjected to micrometre-sized particles and an air humidity level of 45%at a temperature of 90℃.Furthermore,it was noted that the critical reaction energy of high-energetic explosives diminishes with an increase in temperature within 25℃−90℃.In the same vein,as the particle sizes of high-energetic explosives increase,so does the critical reaction energy for micrometre-sized particles.High air humidity significantly affects the sensitivity testing of high-energetic explosives,emphasizing the importance of refraining from conducting sensitivity tests in such conditions.

基于l_(1)/l_(2)正则化的图像盲去模糊方法

针对传统图像盲去模糊方法中出现的图像边缘振铃伪影问题,本文提出了一种基于l_(1)/l_(2)正则化的图像盲去模糊方法。该方法利用清晰图像边缘信息的稀疏性,采用l_(1)/l_(2)范数对未知图像潜影进行约束,对模糊核采用l_(2)范数约束以去除模糊核的噪声。在求解过程中,通过交替迭代的方式更新潜影和模糊核,以估计出的最精细模糊核结合非盲去模糊方法,从而得到去模糊图像。实验结果表明,提出的方法对图像去模糊质量有所提升,有效地去除了一些振铃伪影,对于自然的模糊图像也有较好的复原效果。

General and robust voxel feature learning with Transformer for 3D object detection

The self-attention networks and Transformer have dominated machine translation and natural language processing fields,and shown great potential in image vision tasks such as image classification and object detection.Inspired by the great progress of Transformer,we propose a novel general and robust voxel feature encoder for 3D object detection based on the traditional Transformer.We first investigate the permutation invariance of sequence data of the self-attention and apply it to point cloud processing.Then we construct a voxel feature layer based on the self-attention to adaptively learn local and robust context of a voxel according to the spatial relationship and context information exchanging between all points within the voxel.Lastly,we construct a general voxel feature learning framework with the voxel feature layer as the core for 3D object detection.The voxel feature with Transformer(VFT)can be plugged into any other voxel-based 3D object detection framework easily,and serves as the backbone for voxel feature extractor.Experiments results on the KITTI dataset demonstrate that our method achieves the state-of-the-art performance on 3D object detection.

Automatic detection and removal of static shadows

In case of complex textures,existing static shadow detection and removal algorithms are prone to false detection of the pixels.To solve this problem,a static shadow detection and removal algorithm based on support vector machine(SVM)and region sub-block matching is proposed.Firstly,the original image is segmented into several superpixels,and these superpixels are clustered using mean-shift clustering algorithm in the superpixel sets.Secondly,these features such as color,texture,brightness,intensity and similarity of each area are extracted.These features are used as input of SVM to obtain shadow binary images through training in non-operational state.Thirdly,soft matting is used to smooth the boundary of shadow binary graph.Finally,after finding the best matching sub-block for shadow sub-block in the illumination region based on regional covariance feature and spatial distance,the shadow weighted average factor is introduced to partially correct the sub-block,and the light recovery operator is used to partially light the sub-block.The experimental results show the number of false detection of the pixels is reduced.In addition,it can remove shadows effectively for the image with rich textures and uneven shadows and make a natural transition at the boundary between shadow and light.