Even if image forensics is quite a new topic, some commercial products already exist. Nonetheless, several open problems still exist:
In this work, the Ph.D student will mainly focus on the automatic detection of image manipulations, according to the following three axes:
Currently, tools in image forensics require some manual steps that can be done by trained photo-experts. Automation of such tasks would save time, reduce costs attached to training and facilitate the spread of such tools to larger audiences with some new potential domains of application (e.g. social networks).
Currently, tools work under the assumption that image manipulations are performed to fool human perception, with no assumption on the possible existence of digital detectors. In the close future, one can imagine that attacks will be improved so that they are undetectable by both humans and machines.
We may distinguish between different levels of manipulations:
The current tools are dealing with still images. Nevertheless, videos are more and more important. Extension of algorithms from images to videos is then a priority.
Prospective applicants are encouraged to read the following reference for more information about the exact topic :
Digital image forensics: a booklet for beginners
JA Redi, W Taktak, J.-L. Dugelay J. Multimedia Tools and Applications 51 (1), 133-162
This thesis is granted by the National French Military Research Center.
Education Level / Degree: Computer Engineering or Master Degree (with honors); English is mandatory, French is just a plus.
The application must include (I, II and III):
II-Motivation letter of two pages also presenting the perspectives of research and education
III-Names and addresses of three references
Applications should be submitted by e-mail to //secretariat [at] eurecom [dot] fr" target="_blank">secretariat [at] eurecom [dot] fr with the reference: MM_JLD_PhD_imagerie_042015