3D-TV enhances the viewing experience by introducing depth perception similar to real world scenes. Recent development of Free-Viewpoint Video (FVV) allows the user to interactively navigate between various viewpoints. This is achieved by synthesizing a virtual view via Depth Image Based Rendering (DIBR). In the process of generating virtual views, disocclusion holes appear. These holes are the pixels which were occluded by a closer object in the reference views, but were exposed in the virtual view. To provide a high quality viewer experience these holes should be filled in a visually plausible manner. My research focus is to generate a high quality virtual view by filling the disoccluded hole pixels with appropriate texture and depth information. I am investigating various self-similarity characteristics of an image to help improve the filling process both qualitatively and quantitatively.
My research Interests include Interactive Multi-view Systems, View Synthesis and Free Viewpoint Videos.