DocumentCode :
1779337
Title :
Depth-aware 3D video filtering targetting multiview video plus depth compression
Author :
Aflaki, Payman ; Hannuksela, M.M. ; Homayouni, Majid ; Gabbouj, Moncef
Author_Institution :
Dept. of Signal Process., Tampere Univ. of Technol., Tampere, Finland
fYear :
2014
fDate :
2-4 July 2014
Firstpage :
1
Lastpage :
4
Abstract :
A fundamental challenge of 3D video compression is to provide end users with a high quality video while respecting the limitations of available bandwidth or storage capacities. In this paper, a novel scheme is proposed to filter the texture pictures and depth maps partially targeting a lower bitrate for compression, while preserving the same subjective quality as the one achieved without filtering. In the first step, a depth map is used to calculate a threshold defining the regions of the image where filtering will be applied. This decision is made based on the distance of objects from the camera. The farther the distance of the object, the less likely to have that object be the center of focus for the viewer. In the second step, the content of both texture and depth map belonging to the selected region is filtered. A series of systematic subjective tests showed that the application of such filtering does not decrease the perceived quality of the 3D content compared to the scheme where no filtering is applied. In this case, a BD-rate reduction of up to 17% with an average of 12% is achieved.
Keywords :
filtering theory; video coding; 3D video compression; depth compression; depth map; depth-aware 3D video filtering; multiview video; MVD; Stereoscopic video; low-pass filter; subjective quality assessment;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
3DTV-Conference: The True Vision - Capture, Transmission and Display of 3D Video (3DTV-CON), 2014
Conference_Location :
Budapest
Type :
conf
DOI :
10.1109/3DTV.2014.6874712
Filename :
6874712
Link To Document :
بازگشت