An efficient optimisation scheme for scalable surveillance centric video communications

  • Authors:

  • Naeem Ramzan;Toni Zgaljic;Ebroul Izquierdo

  • Affiliations:

  • School of Electronic Engineering and Computer Science, Queen Mary University of London, Mile End Road, London E1 4NS, United Kingdom;School of Electronic Engineering and Computer Science, Queen Mary University of London, Mile End Road, London E1 4NS, United Kingdom;School of Electronic Engineering and Computer Science, Queen Mary University of London, Mile End Road, London E1 4NS, United Kingdom

  • Venue:
  • Image Communication
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

State-of-the-art coders have been optimised over years according to the needs of the broadcasting industry. There are however key applications of coding technology whose challenges and requirements substantially differ from broadcasting. One of these key applications is surveillance. In this paper an efficient approach for surveillance centric joint source and channel coding is proposed. Contrasting conventional coders, the proposed system has been developed according to the requirements of surveillance application scenarios. It aims at achieving bit-rate optimisation and adaptation of surveillance videos for storing and transmission purposes. In the proposed approach the encoder communicates with a video content analysis (VCA) module that detects events of interests in video captured by CCTV. Bit-rate optimisation and adaptation is achieved by exploiting the scalability properties of the employed codec. Temporal segments containing events relevant to surveillance application are encoded using high spatio-temporal resolution and quality while the portions irrelevant from the surveillance standpoint are encoded at low spatio-temporal resolution and/or quality. Furthermore, the approach jointly optimises the bit allocation between the wavelet-based scalable video coder and forward error correction codes. The forward error correction code is based on the product code constituting of LDPC codes and turbo codes. Turbo codes show good performance at high error rates region but LDPC outperforms turbo codes at low error rates. Therefore, the concatenation of LDPC and TC enhances the performance at both low and high signal-to-noise (SNR) ratios. The proposed approach minimises the distortion of reconstructed video, subject to constraint on the overall transmission bit-rate budget. Experimental results clearly demonstrate the efficiency and suitability of the proposed approach in surveillance applications.