IEEE Signal Processing Chapter Lecture – When Discrete Optimisation Meets Multimedia Security by Dr Shujun Li, University of Surrey.
Selective encryption has been widely used for image and video encryption due to many practical reasons such as to achieve format compliance and perceptual encryption, to avoid negative impact on compression efficiency, and to make the multimedia processing pipeline more modular and thus reconfigurable. The seminar will present research on modelling recovery of missing information with different structures in digital images as a discrete optimisation problem.
In the context of selective encryption, the structure of missing information is defined by the underlying selective encryption algorithm, where the selectively encrypted information is considered missing from an attacker’s point of view. Experimental results showed that the new approach can significantly improve the performance of error-concealment attacks compared to the state of the art in terms of visual quality of the recovered images. The approach can be applied to other areas of multimedia security and multimedia processing in general where the structure of missing information in digital signals is known. An example of adapting the model to self-recovery image authentication watermarking will be shown.