Reducing the Central Fixation Bias in Scene Viewing