Visual-quality evaluation methodology for multiview displays

Multiview displays are characterized by a multitude of parameters, such as spatial resolution, brightness, 3D-crosstalk, and so forth, which individually and in combination influence the visual quality of 3D scenes. These parameters are specified by values precisely measured by optical means. However, it is difficult for an average consumer or content producer to compare the visual quality of two displays or to judge whether given 3D content is suitable for a certain display using only this set of parameter values. In this paper, we propose a quality-measurement methodology that aims to measure the visibility of structural distortions introduced by a multiview display by using a number of test signals with different frequency content and apparent depth. We use these measurements to derive what we call display passband for signals at different disparity levels. The passband determines the frequency components for which the intended signal is predominantly visible with respect to the distortion introduced by the display. Additionally, we propose a method to determine the approximate effective resolution of a display for signals with a given apparent depth. The result of the measurements can be used to compare the perceived visual quality of different multiview displays.

► We propose quality measurement methodology for multiview displays.
► We derive the display passband for signals at different disparity levels.
► The passband determines the frequency components for which the intended signal is predominantly visible.
► We determine the approximate effective resolution for signals with a given disparity.