Efficient Interpolation Architecture Design for Motion Compensation in H.264/AVC
DAI Yu,LI Dong-xiao,ZHENG Wei,LUO Kai,ZHANG Ming
DOI: https://doi.org/10.3785/j.issn.1008-973x.2009.02.012
2009-01-01
Abstract:For decreasing the hardware complexity of motion compensation in the video coding/decoding standard H.264/AVC,an efficient interpolation architecture was proposed,which including a input parallel output architecture for luma interpolation and a chroma interpolation architecture which only needs two multipliers and one adder.The interpolation architecture uses serial Wiener filter for half horizontal samples,selects integer samples by analyzing the data correlation of bilinear sample filter,optimizes the quarter sample filtering,adopts vertical-first interpolation order by analyzing the overlapped region of interpolation windows,and loads small interpolation window according to the motion vector.The simulation shows that the proposed design saves the horizontal 6-tap FIR filters,reduces the size of register array,and saves the memory bandwidth.The proposed design was implemented with 0.18 μm CMOS technology and embedded into a H.264/AVC decoder.Simulation and synthesis show that compared with other designs,this design has lower hardware cost.