With　the　rapid　development　of　Internet　of　Things　and　communication　technologies,　QR　code　images　are　widely　used　in　the　embedded　automatic　identification　field.　However,　previous　works　for　the　location　of　QR　code　have　not　considered　how　to　extract　the　vertexes　accurately　in　geometric　distorted　QR　code　images.　In　this　paper,　utilizing　the　characteristics　of　QR　code,　we　propose　an　effective　mathematical　morphology　method　based　on　self-adapting　structural　element　to　obtain　the　distorted　vertexes　precisely　and　revise　the　distortion.　Our　method　evaluates　two　data　sets.　Compared　with　previous　works,　the　experiments　indicate　that　our　method　can　achieve　higher　recognition　rate　and　be　more　adaptive　for　geometric　distorted　QR　code　images.　Thus,　the　proposed　method　provides　valuable　application　for　embedded　system　and　mobile　terminal　device.　©　2012　IEEE.