This　paper　presents　a　preprocessing　method　for　pedestrians　tracking　and　counting　in　sequences　of　grayscale　images　acquired　by　a　stationary　camera　on　the　bus.　The　objective　is　to　integrate　this　technology　with　a　traffic　control　application　such　as　bus　carrier　traffic　statistics　at　every　station.　The　proposed　method　can　also　be　used　to　detect　and　track　humans　who　will　get　on　or　get　off　the　bus.　Furthermore,　the　proposed　approach　can　be　employed　for　the　multi-object　detection.　Among　these　tasks,　object　segmentation　and　labeling　are　perhaps　the　most　important　problem　in　single　image　processing.　So　the　aim　of　the　paper　is　to　present　the　key　technologies　of　object　segmentation　and　labeling　for　the　tracking　in　video　sequences　taken.