Effect of interlacing methods of stereoscopic displays on perceived image quality

Abstract

To provide two-eyed views with one device, stereoscopic 3D (S3D) displays interlace the two views either temporally or spatially: temporal interlacing (TI) alternates the two views in time with full resolution, while spatial interlacing (SI) presents the two views simultaneously but with half resolution for each eye. We investigate the effect of interlacing methods on image quality through a psychophysical experiment. We compared four experimental conditions: three S3D interlacing methods (TI, SI with raw sampling, and SI with vertical interpolation), and one nonconventional interlacing method (vertical interpolation). The stimuli were 10 natural stereo images presented at nine levels of pixel sizes (0.64, 0.78, 0.89, 1.00, 1.28, 1.55, 1.78, 2.00, and 2.56 arcmin). To test the effect of interlacing methods per se, we provided all the experimental conditions to the subjects using a single experimental setup: a mirror stereoscope. The results show that TI does not degrade the image quality for any pixel size. SI degrades the image quality when the pixel size is relatively large, but the effect of the two SI methods does not differ significantly. Comparison of SI methods against the vertical interpolation method implies that the primary cause of the degradation in image quality for SI methods is the visibility of the interlacing pattern rather than the loss of high-frequency information.