This　paper　proposes　a　mixed　optimization　algorithm　based　on　RBF　neural　network　(RBF)　and　Particle　Swarm　Optimization　(PSO),　which　is　applied　to　the　doorplate　recognition　for　a　mobile　robot.　The　centers　and　widths　of　RBF　neural　network　are　determined　with　self-increasing　clustering　algorithm,　and　the　improved　particle　swarm　optimization　algorithm　is　used　to　optimize　their　distance　from　the　threshold.　Experimental　results　show　that　this　algorithm　has　an　advantage　over　traditional　neural　network　algorithm　in　terms　of　accuracy　recognition　ratio　and　convergence　rate.　Hence,　the　proposed　algorithm　can　meet　the　needs　of　robot　vision　system.　©　2010　IEEE.