A new video watermarking scheme against geometric distortions is proposed, which is most suitable for DCT-encoded compressed video, like MPEG-2, MPEG-4, H.264, etc. To lower the computational cost, 2-D DFT coefficients are constructed directly from the block DCTs by exploiting a fast inter-transformation. Then a watermark is embedded in a RST (rotation, scaling, and translation) invariant domain generated by performing the Fourier-Mellin transform (FMT) on the 2-D DFT magnitudes. For resisting different video format conversion, the watermark detection is performed in the spatial domain. The experimental results show that the proposed scheme is transparent and robust to typical signal processing attacks, lots of commonly geometric distortions including scaling, rotation and some combination of several attacks, and frame dropping.