Geometry　images　are　a　regular　structure　representation　for　3D　meshes,　which　facilitate　the　compression　of　geometry　data　via　the　image-like　structure.　While　geometry　images　are　frequently　used　in　compressing　a　single　geometric　model,　compressing　3D　mesh　sequence　of　different　connectivities　using　geometry　images　is　low　effective　because　the　inter　correlation　of　the　mesh　sequence　is　ignored　when　each　mesh　is　represented　by　a　geometry　image　individually.　We　present　a　novel　compression　scheme　for　3D　mesh　sequence　of　different　connectivities,　using　registration-aware　geometry　images.　We　first　propose　the　thinplate　spline　based　geometry　image　registration　for　preserving　the　inter　correlation　of　3D　mesh　sequence　which　is　ignored　because　of　the　individual　geometry　image　parametrization.　Then　we　use　JPEG2000　for　compressing　the　residuals　of　the　registration-aware　geometry　image　sequence.　Compared　with　the　scheme　coding　geometry　images　using　JPEG2000,　our　scheme　provides　both　promising　objective　and　subjective　qualities　of　reconstruction　meshes　of　geometry　images.　©　2015　by　Binary　Information　Press.