The remote sensing image registration is one of the most important issues of image fusion. Researchers always focus onsearchingadaptive, accurate and fast registration algorithm. A feature matchingalgorithm based on classification is proposed, which uses the correlation of the classified images to realize a precise registration after a coarseregistration.Experimentresults show thefeasibility andadaptability of the method. To bothhomogeneous images andheterogeneous images, theregistrationresults reachsub-pixelprecision.Thecost ofcomputationis farless than that of theconventionalmethodsthroughgray-basedcorrelation orfeature-basedmatching.