The disclosure describes a method of embedding a digital watermark into a video signal using a time based perceptual mask such that the digital watermark is substantially imperceptible in the video signal. A digital watermark embedder computes a time based perceptual mask comprising gain values corresponding to locations within a frame. The gain value for a location in the frame is changed as a function of the change in one or more pixel values at the location over time. The embedder uses the gain values of the time based perceptual mask to control embedding of corresponding elements of a digital watermark signal such that the perceptibility of the elements of the digital watermark signal is reduced in time varying locations of the video signal.