学术报告:Visual Saliency Modelling in Image Quality Assessment
发布时间: 2015/12/21 20:11|
查看数: 5409
Biography
Dr Hantao Liu is an Assistant Professor inSchool of Computer Science and Informatics at Cardiff University (UnitedKingdom). He received the PhD degree from the Delft University of Technology(The Netherlands) in 2011. His PhD thesis focused on Modelling Perceived Qualityfor Imaging Applications. His research interests include visual media qualityassessment, visual attention modelling and applications, visual sceneunderstanding, medical image perception, and human-machine interaction. He isnow serving as the Chair of IEEE MMTC Interest Group on Quality of Experiencefor Multimedia Communications.
Title of Talk: Visual Saliency Modelling in Image QualityAssessment
Abstract: Advances in image qualityassessment have shown the potential added value of including visual attentionaspects in its objective assessment. Numerous models of visual saliency areimplemented and integrated in different image quality metrics (IQMs), but thegain in reliability of the resulting IQMs varies to a large extent. The causesand the trends of this variation would be highly beneficial for furtherimprovement of IQMs, but are not fully understood. In this talk, we willdiscuss about an exhaustive benchmark study, where the capabilities andpotential limitations of computational saliency in IQMs are assessed. We intendto supply practitioners with a useful guidance for applying saliency models inIQMs, in terms of the effect of saliency model dependence, IQM dependence, andimage distortion dependence.