Abstract
The literature of multimedia forensics is mainly dedicated to the analysis of single assets (such as sole image or video files), aiming at individually assessing their authenticity. Different from this, image provenance analysis is devoted to the joint examination of multiple assets, intending to ascertain their history of edits, by evaluating pairwise relationships. Each relationship, thus, expresses the probability of one asset giving rise to the other, through either global or local operations, such as data compression, resizing, color-space modifications, content blurring, and content splicing. The principled combination of these relationships unveils the provenance of the assets, also constituting an important forensic tool for authenticity verification. This chapter introduces the problem of provenance analysis, discussing its importance and delving into the state-of-the-art techniques to solve it.
Original language | English |
---|---|
Title of host publication | Advances in Computer Vision and Pattern Recognition |
Publisher | Springer Science and Business Media Deutschland GmbH |
Pages | 389-432 |
Number of pages | 44 |
DOIs | |
State | Published - 2022 |
Publication series
Name | Advances in Computer Vision and Pattern Recognition |
---|---|
ISSN (Print) | 2191-6586 |
ISSN (Electronic) | 2191-6594 |
Bibliographical note
Publisher Copyright:© 2022, The Author(s).