Subjective evaluation of visual discomfort caused from stereoscopic 3D video using perceptual importance map

In this paper, we propose a visual discomfort evaluation method for stereoscopic 3D video using perceptual importance map. Generally, visual attention is more engaged by salient region having large disparity than the scene's background or other foreground regions. So, we extract perceptual importance map generated from a color based saliency map, an edge based saliency map and a disparity map to assess visual discomfort. Our assessment framework consists of four sequential stages: perceptually important region extraction; motion-in-depth computation; salient object's velocity, depth and viewing time extraction; and visual discomfort computation. Using this framework, we analyze the effects of motion in respect to depth, viewing time and display size while measuring visual discomfort. The experimental results showed that visual discomfort increases in proportion with the speed of motion-in-depth and viewing time. We also found that the viewer's discomfort caused from watching stereoscopic 3D content on the smaller-sized display is higher than with a larger-sized display.