To recognize or not to recognize – A database of encrypted images with subjective recognition ground truth

Heinz Hofbauer,Florent Autrusseau,Andreas Uhl

doi:10.1016/j.ins.2020.11.047

Heinz Hofbauer, Florent Autrusseau + Show 1 more

Open Access

https://doi.org/10.1016/j.ins.2020.11.047

Copy DOI

Abstract

The assessment of very low quality visual data is known to be difficult. In particular, the ability of humans to recognize encrypted visual data is currently impossible to determine computationally. The human vision research community has widely studied some particular topics, such as image quality assessment or the determination of a visibility threshold, while others are still barely researched, specifically visual content recognition. To this day, there does not exist a reliable recognition index that can be employed for such tasks.In order to enable the study of human image content recognition, and in an attempt to propose a corresponding recognizability index, we build a dataset of selectively encrypted images together with subjective ground-truth about their human intelligibility. The methods of acquisition, setup, protocol, outlier detection, are described and we suggest how to calculate a recognition score as well as a recognition threshold. The performance of traditional visual quality indices to predict human visual content recognition is assessed on these data and found to be inapt to estimate recognition of visual content. Contrasting, structure based recognition indices as proposed for this task are shown to represent a promising starting point for further research. To facilitate the creation of a recognition index and to foster further research into human visual content recognition and its relation to the human visual system we will make the database publicly available.

Highlights

The general assessment of the security of encrypted visual data is difficult
Split of the Database for Acquisition: To prevent viewer fatigue we aimed to keep a single session of ground truth acquisition well below the one hour mark, in order to achieve this we split the database into two test sets: SPLIT1 contains the encryption types fake, j2k and j2kne, SPLIT2 is composed of jxr, jpg and H.265
Since the visual quality indices (VQI) produce a numerical score a threshold is required for the classification, we only reported the Matthews correlation coefficient (MCC) at the threshold with the minimum MCC

Summary

Introduction

The general assessment of the security of encrypted visual data is difficult In this context, security is understood as ‘‘visual security”, i.e. the amount of visual information still present in such protected data, without considering cryptanalysis in the sense of analysing the cryptographic strength of the underlying cipher. In most partial/selective encryption schemes, the employed ciphers are beyond any doubt with respect to their cryptographic strength in any case, e.g. AES in some appropriate mode. The goal is to achieve confidentiality for the visual content, or parts thereof, while still maintaining the file format. The latter means that the encrypted file is still usable as the media file it was before encryption. Format compliance means that a standard compliant decoder should be able to decode the encrypted image/video, and produce

Objectives

Results

Conclusion