摘要
Inthispaper,elitistreconstructiongeneticalgorithm(ERGA)basedonMarkovrandomfield(MRF)isintroducedforimagesegmentation.Inthisalgorithm,apopulationofpossiblesolutionsismaintainedateverygeneration,andforeachsolutionafitnessvalueiscalculatedaccordingtoafitnessfunction,whichisconstructedbasedontheMRFpotentialfunctionaccordingtoMetropolisfunctionandBayesianframework.Aftertheimprovedselection,crossoverandmutation,anelitistindividualisrestructuredbasedonthestrategyofrestructuringelitist.ThisprocedureisprocessedtoselectthelocationthatdenotesthelargestMRFpotentialfunctionvalueinthesamelocationofallindividuals.Thealgorithmisstoppedwhenthechangeoffitnessfunctionsbetweentwosequentgenerationsislessthanaspecifiedvalue.Experimentsshowthattheperformanceofthehybridalgorithmisbetterthanthatofsometraditionalalgorithms.
出版日期
2012年01月11日(中国期刊网平台首次上网日期,不代表论文的发表时间)