面向可重构扫描网络的锁定隔离安全结构

为使可重构扫描网络免受未经授权的访问、恶意仪器对传输数据的窜改和嗅探3种安全攻击的影响,提出了锁定隔离安全结构.该结构首先把彼此不具有安全威胁的仪器分成一组,通过控制隔离信号实现组与组之间的单独访问,以防止恶意仪器对传输数据的窜改和嗅探.然后通过使用锁段插入位保护关键的仪器,只有扫描网络中处于特定位置的多个键值被设置成特定值(0,1序列)时锁段插入位才能打开,能加大未经授权访问的难度.此外,为解决仪器分组多导致硬件开销大和布线困难的问题,提出了仪器分组算法,根据仪器间的安全关系构建无向图,然后对无向图求极大独立集,能有效地减少仪器分组数.在ITC 02基准电路上的实验结果表明,与国际上同类方法相比,所提的安全结构打开锁段插入位所需要的时间增大了7倍,在面积、功耗和布线上的平均减少百分比分别为3.81%,9.02%和4.55%.     

基于稳健设计的SGNS算法的超参数调优方法

通常给定超参数的若干取值选取性能最大的为最优组合(称为直接选优法),但是此方法的稳健性差。为此,提出了一种基于稳健设计的超参数调优方法(称为稳健调优法)。具体地,以SGNS算法中的超参数调优为例,在词语推断任务上实验并得出:经方差分析得到SGNS算法中的七个超参数中的五个对算法预测性能有显著影响,确定为主控因子,其余两个确定为噪声因子,且主控因子中有三个对性能估计的方差有显著影响,因此,调优中仅从期望最大来直接选优是不合理的;稳健调优法与直接选优法两者在预测性能上没有显著差异,但稳健调优法对噪声因子具有较好的稳健性。稳健调优法对一般的深度神经网络的调参有实际的借鉴意义。     

基于自编码和知识蒸馏的表面缺陷检测方法

针对传统的表面缺陷检测方法只能对具有高对比度或低噪声的明显缺陷轮廓进行检测的问题,提出了一种基于自编码和知识蒸馏的表面缺陷检测方法来准确定位和分类从实际工业环境捕获的输入图像中出现的缺陷。首先,设计了一种级联自动编码器（CAE）架构用于分割和定位缺陷,其目的是将输入的原始图像转换为基于CAE的预测蒙版;其次,利用阈值模块对预测结果进行二值化以获得准确的缺陷轮廓;然后,把缺陷区域检测器提取并裁剪出来的缺陷区域视为下一个模块的输入;最后,将CAE分割结果的缺陷区域通过知识蒸馏进行类别分类。实验结果表明,与其他几种表面缺陷检测方法相比,所提出的方法综合性能最好,其缺陷检测平均准确率为97.00%。该方法能够有效地对较小的、边缘不清晰的缺陷进行分割,满足对物品表面缺陷实时分割检测的工程要求。     

Fast method for SAR echo simulation of a three-dimensional ground scene

Due to huge calculation of SAR echo simulation of a three-dimensional (3D) ground scene, a fast simulation method is proposed. First, the data of a 3D ground scene are fractionally interpolated based on the Fractional Brownian Motion (FBM) model. Second, the data interpolated are divided into lots of small facets whose precision meets the simulation requirement, the backscattering coefficients of the small facets are calculated. Third, the shaded area is judged by the comparison method of overlook angles. Three parallel levels are analyzed and the kernel function is designed. Finally, the computing method of reduction adding and external thread are also utilized under the framework of GPU in order to get high efficiency. With the methods mentioned above, the imaging result of experimental DEM data verifies the validity and superiority of the proposed method.     

一种基于DaVinci DM 6446的非标准VGA接口设计与实现*

针对DaVinci数字媒体处理器应用系统中525/625P视频图像在VGA显示器上的显示方法,利用DaVinci SOC处理器视频处理子系统（VPSS）中视频处理后端（VPBE）硬件,完成了一种非标准VGA接口硬件和驱动程序软件的设计与实现,获得更好的视频显示质量,并扩展了DaVinci数字媒体应用系统的显示终端类型。     

Infrared heat treatment of Ti-6Al-4V with electroplated Cu

The objective of this study is to investigate an innovative infrared (IR) technique to enhance adhesion of electroplated copper (Cu) on Ti-6Al-4V without dichromate dipping. The ultimate goal is to develop a Cu coating process on Ti-6Al-4V without hazardous hexavalent chromium (Cr) solution treatments. Cu coatings of around 50 μm were electroplated on Ti-6Al-4V specimens at a current density of 0.03 A/cm² in an acidic Cu solution. To improve adhesion of coatings, IR heat treatments were performed on the Cu-coated samples at different temperatures and durations: 860 °C for 600 s and 875 °C for 20–120 s. This process was accomplished in an attempt to replace the use of dichromate dipping before electroplating. For samples heat treated at 860 °C, no bonding existed, even after 600 s. It is believed that solid-state diffusion prevailed at 860 °C and that 600 s was not enough for sufficient diffusion to occur. Adhesion was poor when samples were heat treated at 875 °C for 20 s. Excellent adhesion was observed when the heat treatment holding time was increased to 40 s. For 90 s, the surface appearance of coatings partially changed from Cu-colored to a grayish color. There was no Cu left on the surface after a 120 s heat treatment. From optical microscopic observations on sample cross sections, an interlayer between the Cu and Ti-6Al-4V formed when heat treated at 875 °C for 40 s and longer. The interlayer thickness increased as the holding time increased, until depletion of Cu. The sheet resistivity of coated specimens was on the order of pure Cu for samples heat treated at 875 °C and less than 90 s. During the 875 °C heat treatment, the following occurred: solid-state diffusion of Cu in Ti-6Al-4V, formation of eutectic solutions, dissolution of Cu and Ti-6Al-4V into the liquid phase, and the formation of intermetallic compounds. The lowest eutectic temperature of 875 °C played a key role in this innovative process of Cu coating on Ti-6Al-4V. This paper was presented at the 2nd International Surface Engineering Congress sponsored by ASM International, on September 15–17, 2003, in Indianapolis, Indiana and appears on pp. 403–10 of the Proceedings.     

"THERMO": COMPUTER PROGRAM FOR SIMULATION OF SURFACE PULSE HEATING

1.IntroductionTheoneofthemostimportantparametersofthethermalcycleduringsurfacepulseheatingistherateofheatingandcooling.Knowledgeofthesevaluesinparticulartemperatureintervalshelpstoprognosticatephasestructureintheheataffectedzones(HAZ),toanalyzecausesofthestructuretransformations.Todaythereisavoluminousliteratureoncalculationofthethermalconditionsforasolidduringsurfaceprocessing.Butthetemperaturevariationsareusuallyexaminedasafunctionoftimeorspacecoordinates.Sometimesthefirstderivativeofthete…     

基于Web服务的物联接入系统设计

由于物联网接入设备与应用业务的多元性与复杂性的增加,实际物联网业务开发中需要对业务与设备做协同处理。针对此类问题,提出一种面向服务的模型架构,并据此以Web服务模式触发各类物联接入设备信息的消息管道流程,并在基础性接入信息处理过程中解决不必要的异构问题。基于Intel复杂指令集的嵌入式开发环境,利用多种语言协作设计实现一个物联接入系统。经实测可顺利完成应用层直接操控感知层的业务功能。     

基于改进的图像局部区域相似度学习架构的图像特征匹配技术研究

该文在AdaBoost算法的基础上提出了一种图像局部区域相似度的学习架构,利用该架构训练图像局部特征来获得低维数、独特的特征描述子,以实现对图像局部区域高精度地匹配.所提学习架构通过学习图像局部区域相似性得到一组非线性弱学习器对图像局部特征进行描述;同时,在响应函数组合形式和弱学习器权重优化配置方面,针对浮点描述子和二值描述子分别提出了新的补丁相似性度量函数作为目标函数的核函数,提高了图像特征相似性匹配效果.该学习架构不会受限于任何预定义的图像特征信息采集模式,能产生基于灰度信息或方向梯度信息的特征描述子.实验结果表明采用这种学习架构获得的特征描述子,在所有对比描述子中图像局部匹配查准率是最好的.所提学习框架能有效地配置优化描述子弱学习器,能提高图像特征描述子对图像尺度和视角变化的鲁棒性.