Three-dimensional　(3D)　segmentation　of　maize　ear　is　the　fundamental　research　of　3D　ear　phenotype.　A　method　of　3D　segmentation　for　maize　ear　was　proposed　in　this　paper.　Firstly,　3D　point　cloud　of　maize　ear　was　acquired　by　using　SmartScan　3D　scanner.　Then　the　point　cloud　was　simplified　by　point　cloud　resampling　to　improve　the　efficiency　of　subsequent　algorithm.　And　a　contraction　transformation　was　employed　to　increase　the　distance　of　neighboring　grains　in　Euclidean　space　by　estimating　the　normal　of　each　point　through　k-neighbor　algorithm.　Finally,　the　point　cloud　of　maize　ear　was　segmented　by　Euclidean　clustering.　Results　showed　that　the　segmentation　rate　of　grains　could　achieve　more　than　90%.　Our　method　could　provide　technology　support　for　the　research　of　3D　ear　phenotype　analysis.　©　2017,　The　Editorial　Board　of　Journal　of　System　Simulation.　All　right　reserved.