• DocumentCode
    3418525
  • Title

    Identification of image fragments for file carving

  • Author

    Al-Sadi, Azzat ; Bin Yahya, Muhammad ; Almulhem, Ahmad

  • Author_Institution
    Comput. Eng. Dept., King Fahd Univ. of Pet. & Miner., Dhahran, Saudi Arabia
  • fYear
    2013
  • fDate
    9-12 Dec. 2013
  • Firstpage
    151
  • Lastpage
    155
  • Abstract
    Recovering images intact is an important process in digital forensics, as they may represent primary evidences in crime cases such as child pornography. Due to file syetems´ fragmentation mechanisms, images may be split into several fragments on a physical storage. As such, recovering images fragments and reconstructing the original images embody challenges for file carving tools particularly when the filesystem metadata are not available. In this paper, we propose a method for image fragment identification using a machine learning approach. Our method exploits features in unknown images fragments, and applies various machine learning algorithms to reconstruct the original images by identifying to which particular image a fragment belongs. We provide the details of our methods as well as a validation of its effectiveness.
  • Keywords
    digital forensics; file organisation; image recognition; image reconstruction; learning (artificial intelligence); child pornography; digital forensics; file carving tools; file system fragmentation mechanisms; image fragment identification; image fragment recovery; image reconstruction; machine learning approach; physical storage; History; Image reconstruction; Transform coding; digital forensics; file carving; fragments identification; machine learning;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Internet Security (WorldCIS), 2013 World Congress on
  • Conference_Location
    London
  • Type

    conf

  • DOI
    10.1109/WorldCIS.2013.6751037
  • Filename
    6751037