Identifying deficits of visual security metrics for images

Abstract

Visual security metrics are deterministic measures with the (claimed) ability to assess whether an encryption method for visual data does achieve its defined goal. These metrics are usually developed together with a particular encryption method in order to provide an evaluation of said method based on its visual output. However, visual security metrics themselves are rarely evaluated and the claim to perform as a visual security metric is not tied to the specific encryption method for which they were developed. In this paper, we introduce a methodology for assessing the performance of security metrics based on common media encryption scenarios. We systematically evaluate visual security metrics proposed in the literature, along with conventional image metrics which are frequently used for the same task. We show that they are generally not suitable to perform their claimed task.