The search functionality is under construction.

Author Search Result

[Author] Khairul MUNADI(2hit)

1-2hit
  • Error Protection for JPEG2000-Coded Images and Its Evaluation over OFDM Channel

    Khairul MUNADI  Masayuki KUROSAKI  Kiyoshi NISHIKAWA  Hitoshi KIYA  

     
    LETTER

      Vol:
    E86-A No:8
      Page(s):
    2056-2060

    In this letter, we propose a new error protection technique for JPEG2000-coded images and also present its evaluation over an OFDM channel. The method exploits the layer structure of the JPEG2000 codestream, a data embedding technique and a forward error correcting code. The main header and data in the top layer are duplicated and protected by the error correcting code. These data are then embedded into the bottom layer for error recovery purposes. Our method offers several features: preserves the same codestream structure as the one in the JPEG2000 part 1 standard, provides multilevel error protection, and can be combined with the existing error resilience technique. Hence, the method accommodates the new requirements for wireless JPEG2000 (JPWL/JPEG2000 part 11).

  • Codeblock-Based Error Concealment for JPEG2000 Coded Image Transmission over RTP

    Khairul MUNADI  Masaaki FUJIYOSHI  Kiyoshi NISHIKAWA  Hitoshi KIYA  

     
    PAPER-Digital Signal Processing

      Vol:
    E90-A No:2
      Page(s):
    429-438

    JPEG2000 compression standard considers a block of wavelet coefficients, called codeblock, as the smallest coding unit that being independently entropy-coded. In this paper, we propose a codeblock-based concealment technique for JPEG2000 images to mitigate missing codeblock due to packet loss in network transmission. The proposed method creates a single JPEG2000 codestream from an image that composed of several subsampled versions of the original image and transmit the codestream over a single channel.The technique then substitutes the affected codeblock in a subsampled image with a copy of the corresponding codeblock obtained from other subsampled images. Thus, it does not require an iterative processing, which is time consuming, to construct an estimated version of the lost data. Moreover, it is applicable for a large codeblock size and can be implemented either in wavelet or codestream domain. Simulation results confirm the effectiveness of the proposed method.