Clustering　is　an　important　research　branch　in　the　field　of　data　mining.　Although　there　are　a　large　number　of　clustering　algorithms,　it　is　difficult　to　find　either　a　certain　clustering　algorithm　for　all　the　data　sets　or　a　best　approach　for　a　fixed　data　set.　In　this　paper,　a　diffused　and　emerging　clustering　algorithm　(DECA)　is　proposed,　which　uses　＇similar　energy＇　to　make　up　for　the　shortcomings　of　many　existing　clustering　algorithms　on　similarity,　and　implements　the　emergence　of　classes　with　high　quality.　Many　experiments　demonstrate　our　algorithm＇s　high　accuracy　without　providing　the　number　of　clusters　in　advance,　compared　with　some　typical　clustering　algorithms　such　as　K-means,　Nearest　Neighbor　and　etc.　©　2011　Springer-Verlag　Berlin　Heidelberg.