Efficient point cloud occlusion method for ultra wide-angle computer-generated holograms

Abstract

Holographic Near Eye Displays (HNEDs) are meant to be the next generation of electronic devices that enable projecting three-dimensional images directly to the eyes. This is possible because computer-generated holograms (CGHs), the backbone of HNEDs, encode accurate wavefront information from virtual objects. An immersive experience requires that these CGHs can reproduce large and highly detailed objects. Moreover, an occlusion culling algorithm is necessary to remove back surfaces that do not contribute to the scene, which provides a better sense of reality. Although there is a vast family of occlusion culling methods, none of these, to the best of our knowledge, consider occlusion when calculating CGHs for field of views (FoV) larger than 90° This work proposes a point cloud occlusion culling method for CGHs that projects images with angles larger than 90° Our approach is based on the geometry of the non-pupil Near Eye Display configuration. It is shown that this configuration provides the proper geometrical conditions that can be used for setting fast occlusion culling of unwanted back points. Occlusion culling with our method is carried out with clouds larger than 7-million-point sources and CGHs resolution of 4 K and 8K. It is demonstrated that our method is at least 5 times faster than current solutions. Furthermore, occluded clouds are used for calculating CGHs that are numerically and optically reconstructed. The obtained results confirm that our method provides high-quality occluded clouds, enabling high-quality production of CGHs with large FoV.