Security Assessment of Partially Encrypted Visual Data: Using Iris Recognition as Generic Measure

Abstract

Security assessment of partially encrypted visual data is known to be difficult. We show that a set of known image quality measures turn out not to be good predictors for encryption strength, especially if a different extent of recognisability is present in the data. Iris recognition applied to encrypted sample data is proposed to assess the protection strength of the employed encryption scheme. When choosing the settings as identified in this work, iris recognition performance turns out to be a viable predictor for security of encrypted data, in more consistent manner compared to image quality measures.