Abstract
This paper presents methods of generating an integral 3D image from multiple depth images generated from the multi-view. As a depth image does not include the 3D information at self-occlusion areas, those areas are displayed dark as missing areas, when a depth image is converted into an integral 3D image. Thus, we developed methods to improve the quality of an integral 3D image by completing the missing area using oblique projection images in converting depth images into integral 3D images. Furthermore, we improved the quality of integral 3D images by interpolating the area, where multiple depth images can not complete, using nearest neighbor method to oblique projection images.
