Sparsity　preserving　projections　(SPP)　rooted　in　compressive　sensing　has　already　been　successfully　applied　in　face　recognition.　SPP　is　based　on　subspace　analysis.　The　key　of　methods　based　on　subspace　analysis　is　seeking　a　good　projection　matrix　to　extract　the　features.　It　allows　the　desirable　characteristics　in　the　original　high-dimensional　space　can　be　preserved　in　the　low-dimensional　subspace.　In　SPP,　projection　matrix　is　closely　related　to　sparse　weighting　vector,　so　the　key　of　SPP　is　seeking　a　sparse　weighting　vector　as　well　as　possible　to　describe　the　original　signals.　In　order　to　get　a　better　sparse　weighting　vector　and　gain　a　higher　recognition　rate,　we　proposed　a　new　method　integrating　Radon　transform　and　SPP　(RadonSPP).　In　this　paper,　we　conduct　experiments　on　the　public　BJUT　Eyebrow　Database　to　verify　the　feasibility　and　validity　of　the　proposed　method.　In　addition,　the　experiment　results　show　that　the　method　integrating　Radon　transform　and　SPP　can　gain　a　higher　recognition　rate　than　SPP,　and　we　explain　the　reason.