文本圖像認(rèn)證技術(shù)研究.pdf

1、西北大學(xué)碩士學(xué)位論文文本圖像認(rèn)證技術(shù)研究姓名：杜敏申請學(xué)位級別：碩士專業(yè)：通信與信息系統(tǒng)指導(dǎo)教師：高寶建20100614AbstractInthelate20thcenturysecurityissuesondi西talmultimediabecomeahotresearchintheinformationsecurityfieldAsaspecial虹ndofdigitalmediatextimageplaysadecisiverol

2、einsociallifeInthecurrentfragileintemetsecuritymechanism，therearelotsofsecurityproblemsinusingsomeimportanttextimagesTherefore，itisnecessarytostudytextimageauthenticationindepthinordertoprotecttheseimportanttextsFirstlyt

3、hecurrentresearchstatusabouttextimageauthenticationisintroducedandanalyzedinthefirsttwochaptersofthispaperThenbinarytextimagesauthenticationandrobustimageauthenticationisstudiedindepthrespectivelyinthelastthreechaptersTh

4、emainresultstotheresearchareasfollows：(1)Becausecurrentbinarytextimageauthenticationalgorithmcannoteffectivelysolvetheproblemthattamperingauthenticationtotheuniformityofimage，anewbinarytextimageauthenticationalgorithmbas

5、edonshortaigi“signatureisproposedInthisnewalgorithm，thedistributionoftheuniformareaintheimageisextractedtoashortsignature，whichCaneffectivelysolvetheissuetamperingauthenticationtotheuniformityofimageWithgoodinvisibilityt

6、healgorithmhasthetheoreticalprobabilityoftamperdetex‘tion75％tothenonuniformareaoftheimage(2)ThroughobservingandanalyzingitisfoundthattumingthecentrepixelofThe16BlocksdoesnotchangethenumberofrunstotherOWorcolumnSCanvector

7、softhebinarytextimageBasedonthisfeature，anewbinarytextimageauthenticationalgorithmbasedonwatermarkisdesigned，bywhichtheuniformareaoftheimageCanbecertifiedandhighaccuracypositioningcanbeachievedTheexperimentsshowthattheal

8、gorithmgalleffectivelycertificatetheuniformareasoftheimage，謝tllahighertamperingdetectionrateandgoodinvisibility(3)ConsideringthecharacteristicsofthecharacterstructureanewtextimagefeatureextractingmethodisdesignedExperime