Transrating-assisted spatial resolution reducing for MPEG-4 video streams

Digital video resizing is paramount for promising wireless video applications because different kinds of mobile terminal devices may have different screen sizes. However, it is difficult to efficiently reuse the motion vectors and macroblock modes for the traditional transcoding schemes of spatial resolution reducing because of the different resolutions between the decoder and the encoder. This paper proposed a novel transrating-assisted spatial resolution reducing scheme for MPEG-4 video streams. MPEG-4 video streams are firstly transcoded to the version of a relative low bitrate. A tactic which proactively sets the high frequency discrete consine transform (DCT) coefficients to zeros is used during this transrating process to implement subsampling implicitly in DCT domain, while keeping the spatial resolution invariant during the transrating process. It leverages the processing advantages in both frequency and spatial domains to facilitate efficient using of motion vectors and macroblock modes while reducing the computational load. The transrated video streams are then downsampled at the receiver to reduce the spatial resolution according to its screen size. Experimental results show that this method provides a better video quality, about 0.1dB-5dB, compared to the traditional spatial resolution transcoding methods, which shows widely promising applications for wireless video communications.