欢迎来到天天文库
浏览记录
ID:36469108
大小:2.46 MB
页数:80页
时间:2019-05-11
《空中动态目标识别技术研究》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库。
1、东北大学硕士学位论文空中动态目标识别技术研究姓名:白闻多申请学位级别:硕士专业:计算机应用技术指导教师:杨光红20060101东北大学硕士学位论文A}'slr∞}StudyonAerialDynamicTargetRecognitionAbstractWiththedevelopingofimagetheory,digitalimageprocessingiswidelyusedineverydepartmentofcountryeconomy.Asallimportantresearchfieldofcomputerscience、digi
2、talimageprocessingnowhasgainbroadattention.Digitalimageprocessingtechnologyhasbeenappliedtotheimageofremotesensing。characteridentification.medicaIimageprocessing,multimediatechniques,imagedatabase,industrialdetection,militaryaflhirsmidsoon.Withthedevelopmentofinterdisciplin
3、aryresearchandtheimprovementofcomputerperformance,imageprocessingtechniqueswillbewidelyusedinmanycomplexapplicationfields.Whilethedesireofmodemwarisbecomingmoreandmoreseverely,itismoreandmoreimportanttorecognizeaircrafts.Thisthesisworksmainlyonthepreprocessingandrecognition
4、oftheimage.Themethodintheapplicationofaerialtargetrecognitionhasbeensimulated.Firstly,theinputimageispreprocessed,includingimagesmoothnessandfiltering,edgedetectionandbinarization.Robert.Sobel.Prewitt.Gauss—Laplaeianandwaveletmethodsarcusedtodetectedgesandtheimageisbinaried
5、.Secondly,momentfeatureoftheimageisextracted.InthispapeLHumomentfeatureofaircraftisextracted.Invafiantmomentinvariantseigenvaluealgorithmisamethodwhichcansolvetheproblemofimagegeometricdistortionandhasnowbeenappliedtothepatternrecognition,imagematchingandmanyotherfieldsofte
6、chnology.Thirdly,momentfeatureisinputtotheartificialneuralnetworkforpatternrecognitionandclassification.Inthispaper,improvedback-propagationnetworkandKohonenneuralnetworkgroupclassifieraleproposed.Finally,basedonthemethodsproposedinthispaper,softwaresystembasedonMATLABisdes
7、igned.Simulationresultsindicatethatthissystemcanrecognizetheaircrafisforsimulation.Keywords:imageprocessing;paRernrecognition;preprocessing;edgedetection;binarization;momentfeatureextraction;neuralnetwork—Ul—独创声明本人声明所呈交的学位论文是在导师的指导下完成的。论文中取得的研究成果除加以标注和致谢的地方外,不包含其他人已经发表或撰写过的
8、研究成果,也不包括本人为获得其他学位而使用过的材料。与我一同工作的同志对本研究所做的任何贡献均已在论文中作了明确的说明并表示诚挚的谢意。学位论文版权使用授权书本学位
此文档下载收益归作者所有