• DocumentCode
    1699283
  • Title

    Hierarchical preprocessing for improving the blotch detection performances

  • Author

    Ammar-Badri, H. ; Benazza-Benyahia, A.

  • Author_Institution
    COSIM Lab., Carthage Univ., Tunis, Tunisia
  • fYear
    2013
  • Abstract
    In this work, we are interested in detecting a special type of degradation called blotch in old films. More precisely, our goal is to improve the detection performances by simultaneously increasing the correct detection rate and reducing the false alarmrate achieved by some conventional detectors. The detection procedure consists of mainly two steps. The regions that are likely to be corrupted are firstly located. Then, a conventional detector is applied only on these candidate regions to locate the blotched pixels. The novelty of our contribution is to detect the candidate regions at different resolution levels. This allows to handle both the different blotch sizes and the varying motion magnitudes of the objects. Experimental simulations carried out on real degraded sequences indicate the efficiency of the proposed approach.
  • Keywords
    history; image motion analysis; image restoration; image sequences; object detection; blotch detection performance; blotch size; cinematic heritage preservation; degraded sequences; detection rate; false alarm rate reduction; hierarchical preprocessing; motion magnitude; old films; Abstracts; Detectors; Subspace constraints; blotch detection; hierarchical representation; motion estimation; outliers;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Signal Processing and Information Technology(ISSPIT), 2013 IEEE International Symposium on
  • Conference_Location
    Athens
  • Type

    conf

  • DOI
    10.1109/ISSPIT.2013.6781886
  • Filename
    6781886