Hrm… youcould put accelerometers on two sides (likeleft side andbottomofthecamera)to getanidea of how much it acceleratesin up/down andleft/right directions (youcan put one onthe backtoofor forward/back).Then using that and some calculus youcouldfind out how much it actually moved.
Anothermethod, that probablywouldn’t be as accurate,would betofilm a special “scale” whichis a seriesof verticaland horizontal lines.Then you “measure” (count) how many lines you “move”inthe videofootage.Depending on yourcamera’sresolution and how fine youcanmakethelines, youcan get decently accurate,but probably notas accurateas the accelerometers. Since you’ll be using the camera’s footage in this suggestion, you may want to do both with the stabilization off and on to see the effectiveness of the optical or digital stabilization.
Athird way, that I’m not sure how wellitwouldwork,isto attach some lightweight rods (wooden dowelswould probably work well) onthecamera in thesame kindsof places asthe accelerometersImentionedearlier.Then using scales (rulers might evenwork wellenough for these scales)measure how muchthe dowels move along the scales (the scaleswould be measuringthe furthestpoint away fromthecamera onthe dowels).
Thoseare myinitial thoughts…there’sprobablya muchsimpler way…
