Illicit uses of images have dramatically risen with deepfake content manipulation or unauthorized access. Securing images at their source i.e., at the image sensor level, is key to addressing the challenges of this field of cybersecurity. The "trusted imagers" concept addresses the need to ensure image security, authentication, and encryption starting at the point of acquisition.
Building on our initial research, notably regarding the in-situ generation of keys, your PhD thesis will focus on finding innovative solutions to integrate security functions into image sensors with the challenge of meeting the requirements of low power consumption and compact integrated architecture, while keeping a high level of security. After an initial phase aiming at the development of the skills specific to the thesis, and depending on your background and interests, your work will involve:
- Developing encryption and/or watermarking algorithms in Python to evaluate their complexity, then proposing compact versions compatible with integration into image sensors.
- Evaluating the impact of algorithmic choices and hardware implementation on image quality.
- Designing and validating hardware architectures that implement the algorithms.
- Designing the integrated circuits implementing these functions.
With the ultimate goal of fabricating an integrated circuit, the work will be conducted at CEA-Leti, using professional IC design tools and software development environments.