To　investigate　the　influence　of　traffic　dynamics　of　the　signalized　intersection　on　car-following　behavior,　a　novel　model　considering　the　driver’s　estimation　of　traffic　dynamics　is　proposed.　Firstly,　two　patterns　of　car-following　behavior　under　different　traffic　conditions　are　derived　according　to　the　two-dimensional　distribution　characteristics　of　acceleration　and　velocity　difference　processed　from　high-precision　vehicle　trajectory　data　extracted　from　UAV　footage.　Secondly,　two　parameters　are　introduced　to　the　full　velocity　difference　model　(FVDM)　to　describe　the　driver’s　perception,　estimate　the　traffic　dynamics　at　signalized　intersections,　and　propose　an　extended　car-following　model　at　signalized　intersections　(SI-FVDM).　Based　on　the　linear　stability　analysis　theory,　the　stability　conditions　of　the　SI-FVDM　are　derived.　Finally,　the　parameters　of　the　original　and　extended　models　are　calibrated　and　verified　through　simulation　experiments.　The　results　show　that　the　SI-FVDM　has　a　larger　stability　regime　than　the　FVDM,　and　the　simulation　results　are　consistent　with　the　theoretical　stability　conditions.　©　ASCE.