Detecting manipulations of digital video stream data

The invention relates to a method of detecting manipulations of digital video stream data, wherein the video stream data represents a sequence (17) of video images comprising at least one moving object that moves relatively to other objects and/or relatively to a background scenery and wherein the method comprises: a) detecting the at least one moving object from the video stream data, b) identifying (19) a kinematic model (21) which describes the movement of the moving object, c) determining (23) deviations of the movement which is performed by the moving object according to the video stream data and of a modelled movement which should have been performed by the moving object according to the kinematic model (21), d) deciding (25) if the deviations indicate a manipulation of the video stream data. In particular, the sequence (17) of video images may be obtained by decompressing (15) compressed video stream data.