Variable and constant bitrate in a DVC to H.264/AVC transcoder

  • Authors:
  • Alberto Corrales-García;José Luis Martínez;Gerardo Fernández-Escribano;Francisco José Quiles

  • Affiliations:
  • Instituto de Investigación en Informática de Albacete, University of Castilla-La-Mancha, Campus Universitario s/n, 02071 Albacete, Spain;Architecture and Technology of Computing Systems Group, Complutense University, Ciudad Universitaria s/n, 28040 Madrid, Spain;Instituto de Investigación en Informática de Albacete, University of Castilla-La-Mancha, Campus Universitario s/n, 02071 Albacete, Spain;Instituto de Investigación en Informática de Albacete, University of Castilla-La-Mancha, Campus Universitario s/n, 02071 Albacete, Spain

  • Venue:
  • Image Communication
  • Year:
  • 2011

Quantified Score

Hi-index 0.02

Visualization

Abstract

Mobile-to-mobile video communications constitute one of the main research areas dealing with the dynamic adaptation of traffic generated by video sources. In a framework where one mobile device sends video information to another, both transmitter and receiver should employ video encoders and decoders with low complexity. In this paper, a Variable/Constant Bitrate DVC to H.264/AVC Transcoder is proposed which takes the advantage of both paradigms in terms of low-complexity algorithms on the end-user device side (DVC encoder and H.264/AVC decoder). The proposed transcoder is based on the hypothesis that common DVC GOPs can be converted to H.264/AVC GOPs without significant rate-distortion and bitrate losses, in a flexible way. An in-depth study of the different frame types available in DVC has been carried out in order to exploit the correlation between them and the most suitable GOP pattern in H.264/AVC. Moreover, a dynamic motion estimation technique is proposed in this paper for optimizing the search area for the motion vectors, with the purpose of being used in combination with the GOP mapping approach. Simulation results show that the proposed approaches reduce the DVC to H.264/AVC transcoder complexity by up to 60% on average, while maintaining the coding efficiency in CBR and VBR scenarios, achieving very high quality results over different types of metrics (both objective and subjective). Finally, we conduct a comparative study with all the most prominent DVC transcoding proposals available in the literature, showing that the proposed transcoder achieves the best results (in terms of PSNR and bitrate).