NewDetectors forWatermarks withUnknown PowerBasedonStudent-t ImagePriors

'1~~~~~~~~~~nference viatheexpJectation maximization (EM)algorithm iS Student-t density function, suchmodelsarespatially adaptiveusedtoinfer thehidden variables ofthis modelandestimate andtheExpectation-Maximization algorithm canbeusedto itsparameters. These aresubsequently usedinthedetectors obtain maximumlikelihood estimates oftheir parameters. Using thatare roposed herein. Receiver operatin characteristic theseimagemodelsdetectors basedon theGeneralized p p p g (ROC)curves areusedbasedon100images fromawell Likelihood RatioandRaotests arederived forthisproblem. knowndatabasetocomparetheperformance oftheproposed Numerical experiments arepresented thatdemonstrate the detectorsto previous detectors. properties ofthese detectors andcompared themwithpreviously proposed detectors. Therestofthis paper isorganized asfollows. Insection IIthe image modelandtheEM algorithm that isusedaredescribed.