|
|
自动化学报 2005
基于目标区域特性的海面小目标检测快速算法, PP. 427-433 Keywords: Smalltargetdetection,averagegrayabsolutedifferencemaximum,computationalcomplexity,signal-to-noiseratio(SNR),ROCcurve Abstract: ？Afastobjectdetectionmethodbasedonobjectregiondissimilarityand1-DAGADM(onedimensionalaveragegrayabsolutedifferencemaximum)betweenobjectandbackgroundisproposedforreal-timedefectionofsmalloffshoretargets.Thencomputationalcomplexity,antinoiseperformance,thesignal-to-noiseratio(SNR)gainbetweenoriginalimagesandtheirresultsasafunctionofSNRoforiginalimagesandreceiveroperatingcharacteristic(ROC)curveareanalyzedandcomparedwiththoseexistingmethodsofsmalltargetdetectionsuchastwodimensionalaveragegrayabsolutedifferencemaximum(2-DAGADM),mediancontrastfilteralgorithmandmulti-levelfilteralgorithm.ExperimentalresultsandtheoreticalanalysishaveshownthattheproposedmethodhasfasterspeedandmoreadaptabilitytosmallobjectshapeandalsoyieldsimprovedSNRperformance.
|
