简介：SKiN技术是一个革命性的技术进步。它通过无绑定线封装技术平台增强了可靠性，减小了热阻并改进了内部寄生电感。有了SKiN技术，包含新宽禁带材料功率器件的封装技术已经出现，并允许它们被用在中高功率领域�

简介：Thisresearchpresentsanalgorithmforfacedetectionbasedoncolorimagesusingthreemaincomponents:skincolorcharacteristics,haircolorcharacteristics,andadecisionstructurewhichconvertstheobtainedinformationfromskinandhairregionstolabelsforidentifyingtheobjectdependenciesandrejectingmanyoftheincorrectdecisions.Hereweusefacecolorcharacteristicsthathaveagoodresistanceagainstthefacerotationsandexpressions.Thisalgorithmisalsocapableofbeingcombinedwithothermethodsoffacerecognitionineachstagetoimprovethedetection.

简介：在许多机器学习应用程序，数据不是免费的，并且为每个数据项目有测试费用。为节俭的原因，最小化测试的某存在工作尝试花费了并且同时，保存一个给定的决定系统的一个特别性质。在这份报纸，我们指出一个人能负担得起的测试费用在一些应用是有限的。因此，一个人不得不牺牲各自的性质在一项预算下面保留测试费用。形式化这期，我们定义测试费用限制属性减小问题，在优化目的是最小化有条件的信息熵的地方。这个问题是test-cost-sensitive属性减小问题和0-1背囊问题的必要归纳，因此它是更挑战性的。我们基于信息获得和测试费用建议一个启发式的算法处理新问题。算法在四UCI上被测试(加利福尼亚大学-Irvine)有各种各样的测试的数据集花费了背景。试验性的结果显示唯一的userspecified参数的适当设置。

简介：在HSV颜色空间的人的皮肤的特征广泛地基于内容在图象检索的区域被使用。因为它的价值为肤色有一个狭窄的范围，H作为基本识别参数被选择并且当照明紧张或皮肤表面的弯曲正在变化时，能保持稳定。有在HSV颜色空格的照明的变化的参数的规则被学习。当照明紧张被改变时，/2能使稳定的浸透和价值或皮肤表面(S+V)的平均数被变化，这第一被发现，并且(S+V)/2随着肤色变化，但是变化的趋势与H的相反。因此，(S+V)/H能被用作能提高HSV能力认出人的皮肤的一个新识别参数。CLC数字TN911.73

简介：Anovelopticaltestingtarget(OTT)withthreedegreesoffreedom(DOFs)isproposedformeasuringthetrackingperformanceofaphotoelectrictheodolite.Themaincomponents,suchastheazimuthaxissystem,thepitchingaxissystem,thelinearmotionsystem,andthesimulatedtargetgenerator,aredesigned.Furthermore,thelinearmodulewiththeformofalarge-spanandlow-stiffnesscantileverbeamisstrengthenedusinganintegratedoptimizationmethod.Afterthestructureisstrengthened,the1st-ordernaturalfrequencyincreasesfrom14Hzto32Hz.Finally,afunctionaltestisperformedandtheresultsshowthatcomparedwithatraditionalopticaltesttarget,thesimulatedtargetgeneratedbythenoveltargetisadvantageousinpractice.

简介：Alasersimulatorisproposedwhichisbasedonahardware-in-loop(HWIL)simulationtestsystem.Asanimportantdeviceinthesimulationtestsystem,itisusedtosimulatethelaserenergyandthedivergenceangle.ThelasersimulatorsimulatestheattenuationprocessoflaserenergybythecontinuousrotationofGlanprismsandthechangingoflaserspotsizethroughthecontinuousmotionofthelaserbeamexpandingtelescopeonthestraightplatform.Thedynamicattenuationvelocityofthelasersimulatorisupto4dB/s,andthedivergenceangleadjustmentvelocityisupto3mrad/s.Themethodisfeasibleforthesimulationtest.

简介：ThelifeofimpregnatedBa-Wcathodeswithanewconstructionhavebeenevaluatedusinganacceleratedlifetestatthreedifferenttemperatures(1170℃,1130℃,1090℃)andconstantcurrentdensity(2A/cm~2).Accordingtotherelationshipoflifewithoperatingtemperatures,anacceleratedequationhasbeensetup.Thecathodelifeatnormaloperatingtemperatureisdeductedbasedontheacceleratedequation.Theresultsshowthatlifeofthenovelcathodeexceeds190,000houratacurrentdensityof2A/cm~2.

简介：Thereadoutintegratedcircuit(ROIC)technologyisoneofthecriticaltechnologiesintheresearchofaninfraredfocalplanearray(IRFPA).Basedonthevirtualinstrumenttechnology,asystemforparametertestofROICisdevelopedforIRFPA.Thecomplexprogrammablelogicdevice(CPLD)isappliedintothesystemtoincreaseitsflexibility.Withhighreliabilityandprecision,alongwiththeintegratedsoftwareandhardwareenvironment,thesystemcantestallkindsofROICs.

简介：Selectionoftestcasesplaysakeyroleinimprovingtestingefficiency.Black-boxtestingisanimportantwayoftesting,anditsvalidityliesontheselectionoftestcasesinsomesense.Areasonableandeffectivemethodabouttheselectionandgenerationoftestcasesisurgentlyneeded.Thisletterfirstintroducessomeusualmethodsonblack-boxtestcasegeneration,thenproposesanewalgorithmbasedoninterfaceparametersanddiscussesitsproperties,finallyshowstheeffectivenessofthealgorithm.

简介：一条新奇测试途径为在可编程的门数组(FPGA)是的地里互连资源(红外)proposed.In测试途径，在FPGA的红外的SB(开关框)被利用了测试IRs.Furthermore，在FPGA的可配置的逻辑块(CLB)也被采用了提高开车能力和差错红外的位置能被监视红外决定联系SBs.As结果，红外能与最小的配置patterns.In最大地被扫描实验，内部地发达的FPGA测试

简介：Anewmeasurement-basedadmissioncontrolalgorithmtosupportQualityofService(QoS)demandisproposedforsoftreal-timeapplications.Inthealgorithm,admissiontestisperformedacrossMultipleTime-Scales(MTS)toaccuratelycapturetrafficfluctuationonvarioustime-scales.ByapplyingtheQoSrequirementsdirectlytoadmissiontest,theMTSalgorithmcanproperlymeettheQoStargetandmaximizethebandwidthutilization.

简介：

简介：Anoveldesignofatwo-axisfaststeeringmirror(FSM)withpiezoelectricactuatorsisproposedforincoherentlaserbeamcombination.ThemechanicalperformanceoftheFSMistested.Theresultsshowthatthetiltingrangeofthemirrorisabout4mrad,andthe1st-orderresonancefrequencyisabout250Hz.Aself-designedgratingencoderistakenasthesensor,whichensurestheoptimalprecisionof10μrad.Thenovelmechanicaldesigncanmeettherequirementofengineeringinincoherentlaserbeamcombination.

简介：Itisrevealedthatthedynamicstabilityof2-Drecursivecontinuous-discretesystemswithintervalparametersinvolvestheproblemofrobustHurwitz-Schurstabilityofbivariatepolynomialsfamily.ItisprovedthattheHurwitz-Schurstabilityofthedenominatorpolynomialsofthesystemsisnecessaryandsufficientfortheasymptoticstabilityofthe2-Dhybridsystems.The2-Dhybridtransformation,i.e.2-DLaplace-Ztransformation,hasbeenproposedtosolvethestabilityanalysisofthe2-Dcontinuous-discretesystems,togetthe2-Dhybridtransferfunctionsofthesystems.TheedgetestfortheHurwitz-Schurstabilityofintervalbivariatepolynomialsisintroduced.TheHurwitz-Schurstabilityoftheintervalfamilyof2-Dpolynomialscanbeguaranteedbythestabilityofitsfiniteedgepolynomialsofthefamily.Analgorithmaboutthestabilitytestofedgepolynomialsisgiven.

简介：扬声器适应测试正规化(ATnorm)是在文章无关的扬声器确认的广泛地使用的分数正规化的最有效的途径，它选择扬声器有一个额外的开发语料库的适应冒名顶替者队以便提高识别性能。在这份报纸，能在原来的ATnorm上提供全面重要优点的ATnorm的改进实现被介绍。这个方法在说话者采用新奇生气类似大小没有一个额外的发展语料库的适应的队模型选择。它能与原来的ATnorm完成可比较的表演并且中等减少计算复杂性。与保存额外的开发语料库的完整的使用，全面系统性能能显著地被改进。结果在这个方法在系统性能提供重要改进，这被显示出的NIST2006扬声器识别评估数据语料库上被介绍，与相对，14.4%逼近相等的错误率(无论何时)，14.6%逼近总体上获得的决定费用功能(DCF)。

简介：UnderType-IIprogressivelyhybridcensoring,thispaperdiscussesstatisticalinferenceandoptimaldesignonstepstresspartiallyacceleratedlifetestforhybridsysteminpresenceofmaskeddata.ItisassumedthatthelifetimeofthecomponentinhybridsystemsfollowsindependentandidenticalmodifiedWeibulldistributions.Themaximumlikelihoodestimations(MLEs)oftheunknownparameters,accelerationfactorandreliabilityindexesarederivedbyusingtheNewton-Raphsonalgorithm.Theasymptoticvariance-covariancematrixandtheapproximateconfidenceintervalsareobtainedbasedonnormalapproximationtotheasymptoticdistributionofMLEsofmodelparameters.Moreover,twobootstrapconfidenceintervalsareconstructedbyusingtheparametricbootstrapmethod.TheoptimaltimeofchangingstresslevelsisdeterminedunderD-optimalityandA-optimalitycriteria.Finally,theMonteCarlosimulationstudyiscarriedouttoillustratetheproposedprocedures.