DCT-Domain Embedded Memory Compression for Hybrid Video Coders

  • Authors:
  • Richard P. Kleihorst;René J. Van Der Vleuten

  • Affiliations:
  • Philips Research Laboratories, Prof. Holstlaan 4, 5656 AA Eindhoven, The Netherlands;Philips Research Laboratories, Prof. Holstlaan 4, 5656 AA Eindhoven, The Netherlands

  • Venue:
  • Journal of VLSI Signal Processing Systems - Special issue on recent advances in the design and implementation of signal processing systems
  • Year:
  • 2000

Quantified Score

Hi-index 0.00

Visualization

Abstract

Hybrid video compression schemes such as MPEG2 and H.263use an image memory for motion-compensated coding. In VLSIimplementations, this image is usually stored in external RAM becauseof its large size. To reduce the overall system costs, we propose tocompress the image by a factor of 4 to 5 before storage, which thenenables embedding of the image memory on the encoder IC itself. Theproposed encoder architecture remains in the DCT-domain, so motionestimation and compensation are now performed from this domain. Tocontrol and guarantee the actual storage, scalable compression isused. A hardware implementation is feasible and worthwhile comparedto traditional encoders with no noticeable loss in performance.