A New Adaptive Filter For Inter-perspective Anti-aliasing

ABSTRACT:By using an auto-stereoscopic3D display doctors can not only get a clearly view of organs, tissues, vessels and position relationship between them, but also can get a motion parallax.But auto-stereoscopic3D displays also have disadvantages. Motion parallax implies a continuous change of perspective. However, auto stereoscopic displays can just have a finite number of views projected into the viewing zone. The change of perspective is smooth or not, which depends on the number of projected views and the angular distance between the views. A too sparse view sampling rate will imply jumps of certain areas of the scene when changing perspective. This jump between tow perspectives called inter-perspective aliasing.In previous literatures display prefilter is used to anti-aliasing which cannot filter all inter-perspective aliasing. A combined filter is proposed by Zwicker which can filter all inter-perspective aliasing. However this combined filter still has disadvantages. For one hand, this filter gets rid of some useful data when filtering aliasing, which will bring excessive blur. For another hand, the filter, which is constant filter, can't adapt itself to all of the images with different depth ranges.In this thesis a new adaptive filter which can adapt different images with different depth ranges has been present to solve these problems. After the comparison and analysis, this new adaptive filter can get higher PSNR and better results than previous filters.