Error-Resilient Scheme for Wavelet Video Codec Using Automatic ROI Detection and Wyner-Ziv Coding Over Packet Erasure Channel

Z. Xue, K.K. Loo, J. Cosmas, M. Tun, L. Feng, P.Y. Yip

    Research output: Contribution to journalArticlepeer-review

    22 Citations (Scopus)
    190 Downloads (Pure)

    Abstract

    The error-resilient for video transmission over the Internet in which regarded as the packet erasure channel is always a tough task and has gained lots of attentions. The main contradictory problem lies between error-resilient and bandwidth usage. Additional redundant data has to be added to achieve robust transmission which leads to huge bandwidth usage. In this paper, an error-resilient scheme called Wyner-Ziv Error-Resilient (WZER) based on a receiver driven layered Wyner-Ziv (WZ) coding framework is proposed. The WZER purposely emphasizes on the protection of the Region of Interest (ROI) area in the frame thus to achieve the better tradeoff between the bandwidth usage and error-resilience. WZER is designed to work for the scenario of wavelet based video coding over packet erasure channel, where several techniques including automatic ROI detection, ROI mask generation, Rate distortion optimization (RDO) quantization, WZ coding with layer design, and packet level Low Density Parity Check (LDPC) code are used. The performances of the proposed WZER are simulated based on average PSNR of luminance, perceptual reconstruction and bandwidth usage and compared with normal Forward Error Correction (FEC) full protection scheme and no protection scheme. The results show the advantages of the proposed WZER over traditional FEC protection, especially in the aspects of the recovery of the subject area and bandwidth efficiency.
    Original languageEnglish
    Pages (from-to)481-493
    JournalIEEE Transactions on Broadcasting
    Volume56
    Issue number4
    DOIs
    Publication statusPublished - 2010

    Fingerprint

    Dive into the research topics of 'Error-Resilient Scheme for Wavelet Video Codec Using Automatic ROI Detection and Wyner-Ziv Coding Over Packet Erasure Channel'. Together they form a unique fingerprint.

    Cite this