In　order　to　solve　the　problem　that　traditional　particle　filter　algorithm　often　collapses　to　a　single　point,　this　paper　proposed　a　new　particle　filter　algorithm　by　adding　the　uniform　sampling　algorithm　to　the　random　resampling.　The　uniform　resampling　algorithm　improved　the　diversity　of　the　particles　in　particle　filter　by　inheriting　the　distributing　bound　of　the　abandoned　particles.　Then,　this　particle　filter　algorithm　with　uniform　resampling　was　used　to　research　the　SINS　non-Gauss　and　non-linear　alignment　problem.　Simulated　data　show　that　the　particle　filter　algorithm　with　uniform　resampling　results　in　more　accurate　alignment　then　the　standard　particle　filter　and　improves　the　stability　of　the　SINS　alignment　algorithm.