Relative importance of depth cues on portable autostereoscopic display

Abstract

Depth of a three-dimensional scene is perceived using several different visual cues. The process of perceiving depth is delicate and vulnerable to artefacts. On autostereoscopic displays the aim is to enhance the sensation of reality of the experience by the added three-dimensionality without the need for aiding glasses. The factors that contribute to achieving a plausible 3D effect on autostereoscopic displays are numerous: for example stereoscopy combined with a supporting set of monocular cues: perspective lines, shadows, and textures.The goal of this study is two-fold. First, to gather knowledge on how different characteristics of a 3D scene impact the correct estimation of depth. Second, how the same characteristics impact on perceived quality. These were studied using subjective tests. The studied characteristics were chosen to be five different depth cues - three monocular and two binocular cues. The chosen monocular cues were shadows, texture and focal depth. For binocular cues, two different stereoscopic disparities were used: one optimal for a portable 3D display, and another designed for a HD resolution and later scaled down for a portable device making the disparity decrease. In addition, combination of all aforementioned cues was examined. All of the cues were also studied with three different compression levels (no compression, JPEG with quality factor of 50, and with quality factor of 90).A quantitative study using a portable autostereoscopic display in controlled laboratory environment was conducted. The results of this study indicate that the stereoscopic depth cues outperform the monocular depth cues in accuracy and speed of estimating depth in a 3D video. Interestingly enough, these results are not consistent with the ratings of perceived quality and acceptance. Furthermore, compressing images seems to have a significant effect on the accuracy and speed of depth estimation, and also on the perceived quality.