A stereoscopic content analysis system with visual discomfort-aware

The behavior of human perception plays an important rule for determining the degree of discomfort of a 3D content. In this paper, we propose stereoscopic content analysis methods with visual discomfort estimation. The system includes a perceptual based content analyzer, a vertical disparity analyzer, crosstalk analyzer, and an anaglyph viewer. The perceptual based content analyzer is composed of dense disparity map estimation using modified random walk framework and attention map estimation using static salient object and spatial-temporal attention (STA) analysis. All these analyzers that give us the flexibility to combine human perceptual factors for stereoscopic content analysis are integrated into one system. The hardware of system is design to process content analysis with full-HD stereo video input at 2fps. We also provide a mobile remote-control application which can control the analyzers and receive the current result of the system.