Advanced incomplete 3D representation of video objects using trilinear warping for novel view synthesis

We present a synthesis method for arbitrary virtual views in a multi-view camera set-up. This method generates a view adaptable reconstruction of a 3-dimensional, (3D), object taken from at least two cameras. In this method, we exploit the recently developed Incomplete 3D, (IC3D), algorithm, a compact disparity-based representation of stereo images. Previously, only virtual views inside the baseline of a parallel stereo rig capturing a mainly convex 3D object could be described by IC3D. Hence, the field of applications in virtual environments (VE) was restricted. Thus, to be able to create virtual views of generic 3D objects for camera positions outside the baseline of a convergent multiview set-up, we extended the original IC3D approach towards a more generalised approach, using rectification, trilinear warping, image-based rendering and hidden layer techniques to synthesise new virtual camera views.