基于目标识别的多传感器图像信息融合技术

本文针对目标识别问题，论述了多传感器目标属性融合技术。包括目标属性融合概念、原理、处理结构、算法及应用举例。指出了多传感器目标属性融合中存在的主要问题及解决方法，并预测了发展趋势     

基于偏振信息探究水下环境气泡群对目标成像的影响

水下光学成像是海底探索和目标识别的一个重要方式.由于海浪、船舶尾流以及海洋生物游动与呼吸等原因,存在着大量的气泡.气泡群的光散射作用往往会使水下目标成像效果受限、难以识别,并且一般的光学技术难以消除气泡对成像的影响.针对上述问题,本文先从理论上推导和仿真了入射光线在水下单气泡、气泡群中以及目标表面的光强和偏振信息的变化;然后在构建了水下气泡实验平台的基础上探究了光源入射角度的改变以及成像波段的变化对气泡环境中目标偏振成像的影响;研究了不同金属材质目标物的强度和偏振信息的变化趋势;分析了水下目标在不同气泡群厚度条件下强度和偏振信息的变化趋势;最后利用偏振特征提取与视觉信息保留的图像融合方法抑制气泡对水下目标成像的影响.实验结果显示气泡群中目标成像会受到多种因素的影响,利用偏振图像融合方法会使气泡群受到较好的抑制,并提高了水下目标的清晰度.     

基于自适应补偿网络的视网膜血管分割

基于眼底图像的视网膜血管精确分割对眼科疾病诊断意义重大。但视网膜血管结构高度复杂,多尺度及前、背景比例失衡,自动分割困难。因此,本文提出自适应补偿网络（SACom）实现端到端的视网膜血管精确分割。SACom以U型网络为基本框架,首先在编码器端引入可变形卷积提高复杂血管结构信息学习能力;然后在U型网络底部设计自适应多尺度对齐上下文模块提取并聚合多尺度上下文信息,对齐上下文特征;最后在解码器端设计协同补偿分支,融合多级输出提升模型的映射能力,实现精细分割。实验结果表明,SACom可有效提高视网膜血管的分割精度,在DRIVE、CHASE＿DB1和STARE三个公共数据集上的准确率分别达到0.9695、0.9763和0.9753,灵敏度分别达到0.8403、0.8748和0.8506,曲线下面积（AUC）分别达到0.9880、0.9917和0.9919。     

可调节自适应遥感影像融合方法研究

大部分常用的遥感影像融合方法都存在一个缺陷:只能产生一个特定的融合结果,用户无法控制最终的结果应该保留多少光谱信息或细节信息。提出了一种基于小波变换的可调节自适应遥感影像融合方法,该方法首先分别将待融合影像进行小波分解,然后,通过引入2个可调节参量,在小波域内融合,最后通过小波逆变换得到融合结果。使用法国地球资源探测卫星(SPOT)图像和陆地资源卫星专题绘图仪(landsat TM)图像,将该方法与传统的小波变换融合法、强度色散饱和变换融合法和主成分变换法进行对比试验,结果表明,该方法可以在细节保留和光谱保持两方面达到不同程度的平衡,在合理的参量组合下,融合图像的目视效果和统计指标优于传统融合方法。     

基于自适应算法的光电跟踪转台摩擦补偿方法研究

为了解决低速状态下跟踪转台在摩擦力矩作用下出现低速爬行现象,以某光电跟踪转台为研究对象,根据Lyapunov稳定性理论和自适应控制理论,提出了自适应摩擦补偿方法。通过在经典伺服控制回路中增加摩擦补偿器,以达到抑制低速爬行现象,提高设备低速平稳性能。对比补偿前后转台的输出速度仿真曲线,表明该方法能够有效地抑制低速爬行现象的产生,提高转台的稳定跟踪性能,适合于工程应用。     

基于特征级融合的多波段舰船目标识别方法

针对复杂情况下海上舰船目标单波段特征识别能力不足的问题,研究可见光、中波红外和长波红外三波段特征图像融合技术,重点解决图像融合方法中存在的算法耗时和融合策略选择的问题,提出了一种新的基于区域协方差矩阵的多波段特征级融合方法,针对可见光图像和红外图像分别设计11维和5维特征向量,协方差矩阵可以将多个特征进行融合,既保证了不同目标之间的区别性,同时又减小计算量。该方法首先利用显著性检测,快速定位图像中的目标区域,然后,针对不同波段图像设计的特征向量定义协方差阵的距离计算公式并进行匹配,通过对图像的一次遍历操作获得积分图像,在协方差计算时达到快速计算的目的,最后利用k-阶最近邻算法对多种舰船目标进行分类识别。利用实拍的3 400余张三波段舰船目标图像作为测试数据。实验主要分为两部分,首先对比单波段和三波段融合识别的识别率,验证所提出的融合方法具有更广的应用范围;然后,在计算效率上对比多种传统的像素级方法,验证采用的特征级融合在计算时间上的优势。实验结果表明,该方法可达到95.1%的识别率,单帧计算耗时约为0.5 s,在实时性和检测率方面都有明显提高。     

微尺度沸腾的压力扰动模型

本文从团聚（ｃｌｕｓｔｅｒｉｎｇ）的分子动力学过程出发,研究微尺度沸腾相变的物理机制。通过分析沸腾系统势能变化趋势,指出核化阶段必然伴随团聚体积的增加,从而导致压力扰动的产生。在微尺度条件下,这种压力扰动将可能主导沸腾系统的相交行为,由此导出相变发生的临界尺度判据,在量级上与实验取得很好的一致。     

基于相位信息的边缘检测在目标识别中的应用

在分析利用相位信息进行图像边缘检测原理的基础上,探讨了基于小波域相位信息的边缘检测方法,并将其应用到环境光强变化的目标识别系统中。该算法不需要设定阈值即可对目标边缘进行可靠的检测。实验结果表明,基于相位信息的边缘检测算法受图像的光照和对比度的影响小,在不同光照度的环境下具有良好的目标识别能力     

自适应特征融合和模型更新的相关滤波跟踪

针对复杂场景下单个特征的稳健性差,以及目标存在背景干扰和目标遮挡时跟踪失败的问题,提出一种基于自适应特征融合和模型更新的相关滤波跟踪算法。该算法在核相关滤波的基础上,通过对不同特征的响应图采用平均峰值-相关能量的方法进行加权求和,实现了响应图层面的自适应特征融合。根据响应图的峰值特性计算自适应权重,以其作为置信度确定模型的更新率,进而设计自适应模型更新方法。实验结果表明,该算法能够很好地适应背景干扰、目标遮挡、旋转运动等复杂场景,与近年来优秀的相关滤波跟踪算法相比,所提算法的平均距离精度比其中最优的算法提高了2.64%,平均重叠精度提高了1.54%。     

Symmetrical dynamics of peak current-mode and valley current-mode controlled switching dc-dc converters with ramp compensation

The discrete iterative map models of peak current-mode (PCM) and valley current-mode (VCM) controlled buck converters, boost converters, and buck--boost converters with ramp compensation are established and their dynamical behaviours are investigated by using the operation region, parameter space map, bifurcation diagram, and Lyapunov exponent spectrum. The research results indicate that ramp compensation extends the stable operation range of the PCM controlled switching dc--dc converter to D>0.5 and that of the VCM controlled switching dc--dc converter to D<0.5. Compared with PCM controlled switching dc--dc converters with ramp compensation, VCM controlled switching dc--dc converters with ramp compensation exhibit interesting symmetrical dynamics. Experimental results are given to verify the analysis results in this paper.     

Discrete-time movement model of a group of trains on a rail line with stochastic disturbance

This paper presents a discrete-time model to describe the movements of a group of trains, in which some operational strategies, including traction operation, braking operation and impact of stochastic disturbance, are defined. To show the dynamic characteristics of train traffic flow with stochastic disturbance, some numerical experiments on a railway line are simulated. The computational results show that the discrete-time movement model can well describe the movements of trains on a rail line with the moving-block signalling system. Comparing with the results of no disturbance, it finds that the traffic capacity of the rail line will decrease with the influence of stochastic disturbance. Additionally, the delays incurred by stochastic disturbance can be propagated to the subsequent trains, and then prolong their traversing time on the rail line. It can provide auxiliary information for rescheduling trains when the stochastic disturbance occurs on the railway.     

Performance evaluation of traveling wave ultrasonic motor based on a model with visco-elastic friction layer on stator

A new visco-elastic contact model of traveling wave ultrasonic motor (TWUSM) is proposed. In this model, the rotor is assumed to be rigid body and the friction material on stator teeth surface to be visco-elastic body. Both load characteristics of TWUSM, such as rotation speed, torque and efficiency, and effects of interface parameters between stator and rotor on output characteristic of TWUSM can be calculated and simulated numerically by using MATLAB method based on this model. This model is compared with that one of compliant slider and rigid stator. The results show that this model can obtain bigger stall torque. The simulated results are compared with test results, and found that their load characteristics have good agreement.     

Mode shift and stability control of a current mode controlled buck-boost converter operating in discontinuous conduction mode with ramp compensation

By establishing the discrete iterative mapping model of a current mode controlled buck-boost converter, this paper studies the mechanism of mode shift and stability control of the buck-boost converter operating in discontinuous conduction mode with a ramp compensation current. With the bifurcation diagram, Lyapunov exponent spectrum, time-domain waveform and parameter space map, the performance of the buck-boost converter circuit utilizing a compensating ramp current has been analysed. The obtained results indicate that the system trajectory is weakly chaotic and strongly intermittent under discontinuous conduction mode. By using ramp compensation, the buck-boost converter can shift from discontinuous conduction mode to continuous conduction mode, and effectively operates in the stable period-one region.     

Cylindrical Ising nanowire with crystal field: existence of a dynamic compensation temperatures

We present a study, within a mean-field theory, of the kinetics of the spin-1 Blume–Capel model on cylindrical Ising nanowire in the presence of a time-dependent oscillating external magnetic field. We employ the Glauber transition rates to construct the mean-field dynamical equations. We investigate the thermal behavior of the dynamic order parameters. From these studies, we obtain the dynamic phase transition (DPT) points. Then, we study the temperature dependence of the dynamic total magnetization to find the dynamic compensation points as well as to determine the type of behavior. We also investigate the effect of interaction parameters on the compensation phenomenon and construct the phase diagrams in four different planes. The system exhibits the compensation temperatures, or the N-, P-, Q-, S-type behaviors. Furthermore, we also observed two-compensation temperatures, namely W-type behaviors, whose result is compared with some experimental works and a good overall agreement is found.