SIFT　operator-based　registration　methods　have　been　found　efficient　matches　between　features　with　unique　local　neighborhoods,　yet　they　fail　to　handel　global　context　to　resolve　ambiguities　that　can　occur　locally　when　images　to　be　registered　have　similar　regions.　This　paper　presents　an　novel　image　registration　method,　based　on　local　SIFT　operator　and　cross-correlation　information.　The　cross-correlation　information　can　discriminate　local　features　that　have　similar　local　appearance.　Moreover,　the　proposed　image　registration　algorithm　has　better　robust　performance　in　matching　under　noise　environments.　Simulation　results　confirm　that　the　proposed　algorithm　can　give　better　image　match　than　two　related　image　registration　algorithms　under　noise　environments.　©　2011　IEEE.