Histogram　Intersection　(HI)　kernel　has　been　recently　introduced　for　image　recognition　tasks.　The　HI　kernel　is　proved　to　be　positive　definite　and　thus　can　be　used　in　Support　Vector　Machine　(SVM)　based　recognition　[1].　In　this　paper,　Profile　Histogram　Intersection　(PHI)　kernels　and　Detail　Histogram　Intersection　(DHI)　kernels　function　are　proposed　which　based　on　traditional　Histogram　Intersection　kernels.　In　experiments　on　classification　of　face　images,　our　DHI　kernel　shows　higher　classification　accuracy　than　HI　kernels.　©2009　IEEE.