Female or male observers ? - Gender differences in subjective image quality assessment

Abstract

Objective image quality assessment measures were extensively used to evaluate the performance of different image processing and analysis algorithms. However, they are application-driven. In contrast, subjective assessment measures are general-purpose and considered as the ground truth. Several Subjective Image Quality Assessment (SIQA) methods have been proposed in the literature. These measures require normalized environment and standardized protocols and are mainly based on the visual judgments of image quality. Nevertheless, these judgments are strongly influenced by human factors which might skew the precision of the assessment. Consequently, when selecting observers panel, researchers take into account several criteria such as age, sex, ethnic origin, socio-cultural and professional. Following an in-depth analysis of recent state-of-the-art work, we found that no one has investigated a gender difference in human judgment. In this paper, we aim to highlight the effect of this property in the subjective evaluation tests that are conducted on the TID2008 database. Obtained results show that men and women judge the same distorted images differently, with high precision in favor of women, outperforming in several cases the results provided by the TID2008 database.