Classifying Perceptual Experience of Tone-mapped High Dynamic Range Videos through EEG

Abstract

High dynamic range (HDR) imaging has potential for providing immersive experience of multimedia contents. HDR contents are expected to have better perceptual quality than conventional low dynamic range (LDR) contents, but the perceptual difference in the brain between HDR and LDR contents has not been adequately studied. In this paper, we investigate perceptual experience of tone-mapped HDR videos based on electroencephalography (EEG) classification. A support vector machine (SVM) classification system is constructed using the acquired EEG signals to explore implicitly measured perceptual difference between tone-mapped HDR and LDR videos. As a result, average accuracies of 82.14% and 42.86% are obtained in a subject-dependent scenario and a subject-independent scenario, respectively. This shows that it is possible to distinguish perceptual responses for tone-mapped HDR and LDR videos in a subject-dependent manner. Further, features selected for classification are investigated in each classification scenario. Although the spatial position of the features on the scalp varies across subjects, gamma band powers are generally effective for classification.