DocumentCode :
1258668
Title :
Methodology for design of antialiasing filters for autostereoscopic displays
Author :
Boev, Atanas ; Bregovic, Robert ; Gotchev, Atanas
Author_Institution :
Dept. of Signal Process., Tampere Univ. of Technol., Tampere, Finland
Volume :
5
Issue :
3
fYear :
2011
fDate :
6/1/2011 12:00:00 AM
Firstpage :
333
Lastpage :
343
Abstract :
Multi-view autostereoscopic displays can be modelled as a multirate system. The design of such displays involves a construction compromise between the amount of different views and the spatial resolution of each view. Images to be visualised on these displays are prone to aliasing errors. Careful antialiasing requires knowledge about the display frequency response, which is determined mainly by the view sub-sampling topology but it is also influenced by some other, generally non-linear, aliasing-causing display effects. In this work, a methodology for designing antialiasing filters for autostereoscopic displays is proposed. It includes the following three steps: (i) measuring aliasing effects by a set of test images - displayed on the screen, then photographed and then analysed in Fourier domain; (ii) estimating the display passband based on the set of measurements and (iii) designing filters confined to the so-estimated band. Using this methodology, one non-separable and three separable antialiasing filters have been designed. The non-separable filter cancels the aliasing terms completely, whereas the separable approximations allow for some small amount of aliasing for the sake of perceptually favoured sharpness preservation. The advantage of the methodology with respect to previously suggested antialiasing filter design approaches is demonstrated by objective comparisons of filter performance and computational efficiency and by visual inspection on a set of test images.
Keywords :
spatial filters; three-dimensional displays; aliasing effect; antialiasing filter; autostereoscopic displays; display passband estimation; spatial resolution; view subsampling topology;
fLanguage :
English
Journal_Title :
Signal Processing, IET
Publisher :
iet
ISSN :
1751-9675
Type :
jour
DOI :
10.1049/iet-spr.2010.0039
Filename :
5931032
Link To Document :
بازگشت