The　high　dimensional　data　brings　dimensionality　disaster　problem　in　information　fusion　and　pattern　recognition,　the　similarity　frequency　band　between　feature　information　is　clustered　by　the　improved　fuzzy　clustering　algorithm,　the　feature　extraction　and　data　dimension　reduction　are　realized　in　feature　distributed　band　respectivly.　The　threshold　and　clustering　parametre　are　established　by　compactness　degree　and　separation　degree　between　samples,　the　cluster　validity　criterion　function　is　given　to　determine　the　optimal　clustering　for　samples,　and　finally　an　illustration　verifies　this　method.　©　2010　IEEE.