Paper Info
Title
Foreground-Object-Protected Depth Map Smoothing for DIBR
Abstract
Depth image based rendering (DIBR) uses a 2-D color image and its associated depth map to render virtual views and then a stereoscopic image. One of the main problems in DIBR is how to reduce the size of holes in the generated virtual view image. Smoothing the depth map before image warping is a common solution. However, the previous smoothing methods bring edge distortions to the foreground objects which are more valuable than the background. In this paper, a piecewise smoothing filter is adopted, performing filtering on the depth map only in the background regions while leaving the foreground regions unchanged. And the filtering is also controlled by the registration points of the depth map. Experimental results show that the proposed method obtains better subjective qualities for the virtual views, and it is more time-saving compared to the former methods while having a better PSNR performance.
Year
DOI
Venue
2012
10.1109/ICME.2012.116
ICME
Keywords
Field
DocType
depth map,foreground object,image warping,2d color image,smoothing methods,depth map registration,virtual view image,depth image based rendering,rendering (computer graphics),background region,registration points,stereoscopic image rendering,piecewise smoothing filter,associated depth map,edge detection,2-d color image,stereoscopic image,foreground object protection,dibr,image registration,edge distortions,stereo image processing,virtual view,depth image,foreground-object-protected depth map smoothing,depth map smoothing,virtual view rendering,color,psnr,low pass filters
Computer vision,Image warping,Pattern recognition,Computer science,Edge detection,Stereoscopy,Smoothing,Artificial intelligence,Depth map,Image-based modeling and rendering,Image registration,Color image
Conference
ISSN
ISBN
Citations 
1945-7871
978-1-4673-1659-0
2
PageRank 
References 
Authors
0.40
4
3
Name
Order
Citations
PageRank
Xiao-han Lu120.40
Fang Wei220.40
Fang-min Chen321.08