A　novel　image　annotation　algorithm　based　on　Markov　model　is　proposed　to　bridge　the　semantic　gap　of　content-based　image　retrieval.　This　algorithm　treats　each　candidate　keyword　as　a　state　in　Markov　chain,　and　implements　image　annotation　by　estimating　the　probability　of　Markov　transition.　Compared　with　classical　algorithms,　the　proposed　algorithm　consider　correlation　between　keywords　that　improves　results　of　image　annotation,　Experimental　results　on　the　typical　Corel　dataset　demonstrate　the　effectiveness　and　the　increasing　annotation　precision　of　our　proposed　algorithm.