Efficient Implementation of Interpolation for AVS

  • Authors:
  • Shu Hu;Xiaoxia Zhang;Zhigang Yang

  • Affiliations:
  • -;-;-

  • Venue:
  • CISP '08 Proceedings of the 2008 Congress on Image and Signal Processing, Vol. 3 - Volume 03
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

The framework of current state-of-the-art video coding standard is block transform/block prediction hybrid based scheme. Sub-pixel motion compensation can reduce bit rate and improve the compressing efficiency in video coding, while sub-pixel interpolation can also bring great increment in computational complexity. In this paper, pipelined solutions to the interpolation in AVS are presented based on the digital signal processor (DSP) platform. The ways of optimizing the interpolation part of video encoder are discussed, and many optimization strategies, including algorithm level modification, refining C code, writing linear assembly, using DMA and organizing memory, are proposed to implement the video application system. The final simulated results prove that good DSP performance can be achieved when utilizing all these optimization strategies.