Background　pixels　in　object　model　will　induce　localization　error　of　object　tracking.　To　reduce　the　error,　a　coefficient　based　on　the　color　histograms　of　the　object　pixels　and　background　pixels　around　the　object　is　computed.　Then,　this　coefficient　is　incorporated　into　the　computation　of　spatiograms　(spatial　histograms).　Therefore,　a　mean　shift　object　tracking　algorithm　based　on　background-weighted　spatiograms　is　proposed.　In　addition,　an　update　scheme　of　the　spatiograms　is　presented.　The　experimental　results　show　the　proposed　tracking　algorithm　has　good　localization　precision　of　tracking,　and　is　robust　against　occlusion.