The impact of the MozJPEG compression library on digital image forensics

DegreeMaster
StatusActive
Supervisor(s)Nora Hofer, MA

Description

MozJPEG is a JPEG compression library that aims to reduce file size while preserving perceived image quality. The library improves the rate-distortion tradeoff with trellis optimization, which modifies the Discrete Cosine Transform (DCT) coefficients of an image. A relevant goal in the field of digital image forensics is to detect manipulations (‘photoshopping’). So far, this community has devoted little attention to the implications of using different JPEG compression libraries. The uptake of MozJPEG calls for a revisit of established methods that have been developed and tested before trellis optimization was commonly used.

The objective of the thesis is to re-evaluate selected forensic methods that detect and localize manipulations in images that have been decompressed from JPEG. The student will implement the methods and measure how sensitive they are to the choice of the specific implementation used for JPEG pre-compression. Ideally, potential performance differences can be attributed to the effect of trellis optimization, and the method can be adapted to increase the reliability for pre-compression with MozJPEG.

References

  • Böhme, R. and Kirchner, M. Media Forensics. In S. Katzenbeisser and F. Petitcolas, eds., Information Hiding. Artech House, 2016, pp. 231–259. [PDF]
  • Hofer, N. and Böhme, R. Progressive JPEGs in the Wild: Implications for Information Hiding and Forensics. In Proceedings of the 11th ACM Workshop on Information Hiding and Multimedia Security (IH&MMSEC). ACM, Chicago, IL, 2023, pp. 47–58. [PDF] [Publisher] [Video]