Three Extensions to Subtractive Crosstalk Reduction

Abstract

Stereo displays suffer from crosstalk, an effect that reduces or even inhibits the viewer s ability to correctly fuse stereoscopic images. In this paper, three extensions for improved software crosstalk reduction are introduced. First, we propose a reduction method operating in CIELAB color space to find a perceptually better color match for crosstalk corrected pixels. Second, we introduce a geometry-based reduction method that operates on fused 3D pixels. Finally, a run-time optimization is introduced that avoids the need to process each pixel. We evaluated our CIELAB-based method using the Visible Differences Predictor (VDP). Our results show that we are able to significantly improve crosstalk reduction compared to previously used methods that operate in RGB color space. The combination of our methods provides an improved, real-time software crosstalk reduction framework, applicable to a wider range of scenes, delivering better quality, higher performance, and more flexibility.