Optimized adaptive depth map filtering

Three-Dimensional (3-D) Television (TV) is believed to be the future TV broadcasting technology which then would replace the traditional 2-D TV systems. Future 3-D TV could enhance the visual home entertainment experience in a way that the user can navigate through the scene or switch the viewpoint. Moreover, new technologies such as autostereoscopic multiview displays provide a 3-D perception to the viewer without the need to wear additional glasses. For these technologies, the need arise to generate additional Virtual Views (ViV) of given scenes with different viewpoints. A new view can be synthesized at the receiver side by utilizing Depth Image-based Rendering (DIBR). Although DIBR has many advantages, one of the key challenges is, how to fill uncovered areas caused by disocclusions and incorrect depth values. To handle such disoccluded areas a new depth preprocessing method is proposed, which is applied prior to 3-D warping. Depth discontinuities in the Depth Map (DM) are smoothed to reduce the number and the size of uncovered areas in the ViV. Considering the fact that filtering the whole image introduces strong distortions, we adaptively weight the filter to reduce filter-induced artifacts. The proposed method shows considerable objective and subjective gains compared to the state-of-the-art.