Fast image identitfication using partial files

  The field of Digital Forensics has been facing scalability issues for the last 10 years, resulting in huge backlogs for law enforcement agencies, impeding the course of justice. One reason for this is the performance bottleneck introduced by storage media, with large disk drives potentially requiring an entire day to duplicate.

My work focuses on methods for evaluating digital images in a manner which is both computationally inexpensive, and avoids the need to read the entire file from the hard disk. Fingerprints are derived using compressed domain Content Based Image Retrieval (CBIR) techniques, which are more robust than standard cryptographic hashing methods.

  • Dates:

    2015 to date

  • Qualification:

    Doctorate (PhD)

Project Team

Themes

Research Areas