Digital Signal Processing Reference
In-Depth Information
From the figures, one can see that the subjective comfort disparity range is
4-5 times smaller than the objectively calculated one. Apparently, the influence of
display properties influences the range more than the viewpoint-related parameters.
More information about comparing various parameters of 3D display can be found
4
Visual Optimization by Signal Processing
Signal processing techniques can be used for improving the visual quality of 3D
displays in three ways. If a distortion introduced by the display can be described
as an invertible function, one can pre-process (pre-distort) the image using the
inverse function. In such case, the changes caused by pre-processing would cancel
display distortions, resulting in a clean signal representation without artifacts. Such
process is known as
pre-compensation
and can be used to improve some cases of
pseudoscopy, hyperstereopsis and ghosting. In the case of distortions which cannot
be pre-compensated, a signal processing algorithm can decrease their visibility,
helping mitigate the perceived annoyance of artifacts, thus improving the quality.
Artifact mitigation algorithms are possible for imaging, aliasing and cases of
pronounced crosstalk. Finally, the visibility of some artifacts does not depend purely
on the content but also on observer position, motion and head orientation. Such cases
need real-time algorithms which actively track the observer and process the visual
signal accordingly.
distortions caused by observation angle one needs to know the position of observer
in respect to the display. Most often this is done by using camera-based tracking
and face- or eye-tracking algorithms. Once the observation position is known the
image can be optimized for the calculated angle and distance. Although user-
viewpoint optimization usually work for one observer only. Ghosting artifacts can be
either pre-compensated or mitigated. For dual-view displays, where crosstalk levels
are low, pre-compensation is possible but limits the dynamic range of the display
multiplexed dual-view 3D displays. A similar approach can be used for a multiview
display if a single observer is tracked. However, the possibility of multiple observers
and the pronounced crosstalk between neighboring views make crosstalk mitigation
the preferred approach for multiview 3D displays. Such algorithms aim to reduce the
visibility of ghost images by filtering horizontal high-frequency components of the
image but at the expense of losing image details.
The range of artifacts which are caused by the optical separation layer of
deriving the passband of the display, and prepare a filter which removes image
