Quality-preserved and low-complexity frequency-domain down-sizing method in a video decoder

This work is to explore a frequency-domain down-sizing method which preserves picture quality and consumes low complexity in a video decoder. The conventional frequency-domain down-sizing methods usually yield a drift error owing to frequency-domain coefficient truncation and motion compensation using small reference frames. The proposed method can overcome this drawback by using non-zero coefficient compensation and adaptive block-size decoding. In a Group of Pictures (GOP), an I frame and P frames close to the I frame are decoded at a full frame size and then are performed by downscaling. The motion vectors of the other P frames are pre-decoded first and then are analyzed to understand their referring relationship. Based on this relationship, the blocks which are referred and not referred by the other moving blocks are decoded at full and small block sizes, respectively. Additionally, B frames are decoded at an adaptive block size. The simulation results demonstrate that the proposed method can attain the same picture quality and 11%-29% complexity reduction than the conventional method conducting at a full frame size.