In　this　paper,　we　introduce　a　novel　Local　Gabor　Binary　Pattern　Random　Subspace　Method　(LGBPRSM)　for　wearing-glasses　face　recognition.　It　extracts　the　discriminating　features　from　facial　space　based　on　local-feature　method,　after　that,　it　constructs　multiple　classifiers　by　randomly　sampling　from　the　feature　set　to　gain　more　diversity　between　classifiers　for　efficiently　recognizing　the　faces　with　glasses.　Our　experimental　results　on　FERET　and　Yale　database　prove　the　advantages　of　the　proposed　approach　when　compared　with　other　methods.　©2010　IEEE.