As the volume of multimedia data exchanged over the Internet continues to grow, the problem of preventing the unauthorized duplication of copyright-protected images or the dissemination of undesirable material such as pornographic pictures has become increasingly important. The majority of the images circulated over the Internet are coded in the JPEG format. Thus, the present study proposes a novel digital feature retrieval scheme designed to detect unlawful dissemination of pre-selected JPEG images. In the proposed approach, the packet-level features of the target JPEG image are collected in advance and are stored in a feature database. Thereafter, the features of all the packets passing through a nominated router are compared on-the-fly with those in the feature database in order to detect any transmissions of the target images. The experimental results show that the proposed packet-level forensic scheme is far more efficient than existing application-level schemes. As a result, the proposed scheme provides a viable solution for the background monitoring of JPEG streams over large-scale network environments such as the Internet in order to detect unlawful transmissions and to compile digital evidence of such transmissions for possible future legal proceedings.
All Science Journal Classification (ASJC) codes
- Information Systems
- Computer Networks and Communications