Ineachgeneratedview,theholefillingkernelsearchesforcolored pixels in the hole neighborhood that are in the same line of the hole. It then selects those pixels with the lowest depth values (the background pixels) and fills the hole with a weighted average of those background neighbor pixels. The weight is based on the distance from the pixel to the hole. This approach assumes that background pixels are probably the correct ones to fill disocclusion holes.