EM期望最大化算法用于图像分割

Kmeanssegmentationmethodhasanunderlyingassumptionthateachelementcannotbelongtotwoclustersatthesametime.Sometimes,anelementintransitionareabetweentwoclustersarehardtodefine.Thiselementmaybelongtomultipleclusters,withprobabilitie