基于adaboost算法在网络不良照片识别中的方法研究VIP专享VIP免费

基于adaboost算法在网络不良照片识别中的方法研究摘要：随着科技时代的进步，互联网技术在计算机的广泛使用的给人提供便利的同时，也萌发了很多危害。人像检测技术开始于上世纪60-70时代，经过几十年的研究与应用，已经慢慢完善。但时代的进步让识别技术的要求愈来愈高，检测敏感照片是一项重要识别类工作，所以关于这种技术和理论不断地被人们所关注，渐渐成为研究热点。本文针对不良照片中的感兴趣区域的检测提出了一些新方法，改进训练弱分类器的权值方法，高效率的分辨出不良照片，督促互联网的绿色发展。本论文研究的问题是关于Adaboost算法的在不良照片中的检测，但关于识别与检测有许多客观的非人为的干扰因素存在，现在我们需要的是在存在众多干扰的情况下仍然有一个检测水平高，速度快的算法。这也一直是研究人员努力的方向。所以在此基础上，提出来了Adaboost算法下Haar特征用于检测色情图片。新的算法不仅可以迅速，又可以利用这种特征排除多余的干扰因素，促进了工作的顺利进展。I关键词：不良照片；敏感信息识别；人脸检测；AdaBoost；新特征；特征抽取；误检率；IIAbstract:WiththerapiddevelopmentofInternettechnologyandtheprogressofsociety,theInternetbringsconvenienceandspeedtothepeople,butatthesametime,italsosproutsmanyhazards.Portraitdetectiontechnologybeganinthe60-70eraofthelastcentury.Afterdecadesofresearchandapplication,ithasbeengraduallyimproved.However,withthedevelopmentofthetimes,therequirementofrecognitiontechnologyisgettinghigherandhigher.Detectingsensitivephotosisanimportantrecognitionwork.Therefore,peoplepaymoreattentiontothistechnologyandtheory,andgraduallybecomearesearchhotspot.Inthispaper,weproposesomenewmethodstodetectregionsofinterestinbadphotos,improvetheweightmethodoftrainingweakclassifiers,distinguishbadphotosefficiently,andurgethegreendevelopmentoftheInternet.TheproblemofthispaperisaboutthedetectionofAdaboostalgorithminbadphotos,buttherearemanyobjectiveandnon-humaninterferencefactorsinrecognitionanddetection.Whatweneednowisahigh-levelandfastdetectionalgorithminthepresenceofmanyinterference.Thishasalwaysbeenthedirectionofresearchers'efforts.Soonthisbasis,HaarfeatureisproposedunderAdaboostalgorithmtodetectpornographicimages.Thenewalgorithmnotonlycanquickly,butalsocanusethisfeaturetoeliminateunnecessaryinterferencefactors,whichpromotesthesmoothprogressofthework.IIIKeywords:Badphotos;SensitivePhotoRecognition;Facedetection;Adaboost;Newfeatures;Featureextraction;Noisefactor;IV目录摘要........................................................................................................................IABSTRACT................................................................................................................II目录........................................................................................................................IV第一章绪论............................................................................................................71.1研究背景和意义............................................................................................71.2研究现状......................................................................................................81.3主要研究内容与方法.....................................................................................91.4主要工作安排及各章节内容..........................................................................10第二章基于AdaBoost人脸检测算法的原理与研究..........................................112.1引言...............................................