DIGITALIMAGEFORENSICSFOR IDENTIFYING COMPUTER GENERATED AND DIGITALCAMERA IMAGES
暂无分享,去创建一个
We describe adigital imageforensics technique todistinguish imagescaptured byadigital camera fromcomputer generated images. Ourapproach isbased onthefact that imageacquisition inadigital camera isfundamentally different fromthegenerative algorithms deployed bycomputer generated imagery. Thisdifference iscaptured interms oftheproperties oftheresidual image(pattern noise incase ofdigital camera images) extracted byawavelet based denoising filter.In[1], itisestablished that eachdigital camerahasaunique pattern noise associated withitself. Inaddition, ourresults indicate that thetwotypeofresiduals obtained fromdifferent digital cameraimages andcomputer generated images exhibit somecommon characteristics that isnotpresent intheother typeofimages. This canbeattributed tofundamental differences intheimagegeneration processes that yield thetwotypes ofimages. Ourresults arebased onimages generated bytheMayaand3DStudio Maxsoftware, and various digital camera images.
[1] I. Selesnick,et al. Bivariate shrinkage with local variance estimation , 2002, IEEE Signal Processing Letters.
[2] I. Johnstone,et al. Ideal spatial adaptation by wavelet shrinkage , 1994 .