Virtual reality is a new and rapidly expanding area seeking to immerse the user in completely new surroundings. Total immersion is currently hindered in pre-recorded environments by the user’s inability to move inside the scene. Additional views can be synthesized through depth-image-based-rendering (DIBR), but DIBR causes disocclusion holes ‒ visual artifacts due to spatial regions that were not visible in the reference view. These holes can be filled with traditional inpainting (using known portions of the image to fill the unknown portions), but the additional depth data can be leveraged to produce higher quality, more accurate results.
Depth-image-based-rendering (DIBR) is a popular method of synthesizing new views of scenes using traditional images accompanied by depth maps. However these new view suffer from disocclusion holes . background areas revealed by the viewpoint change which have no image or depth data. The filling of disocclusion holes has been an area of development since 2011. This paper seeks to further develop one of the fundamental methods in disocclusion filling by adding small details from more recent (and much more complex) methods.