Boundary Filtering on Synthesized Views of 3D Video

When we generate an intermediate viewpoint image using multi-view images and their depth maps, we may have very annoying boundary noises in the background due to depth value errors around depth discontinuities. In this paper, we propose a boundary filtering method for synthesized images. After defining the boundary noise area using depth map, we replace them with corresponding texture information from the other reference image. Our experiments show that the proposed method improves visual quality of the synthesized virtual viewpoint images.