Computer-Generated Holography With Deep Learning

Most mainstream commercial solutions for 3D display are based on binocular vision principles to achieve more realistic visual experiences. However, unlike when viewing real 3D objects, the viewer’s depth of visual focus remains constant while wearing the device to obtain 3D information. This vergence accommodation conflict predisposes the viewer to visual fatigue and vertigo, limiting the user’s experience. Computer-generated holography (CGH) can prevent the origin of vergence accommodation conflict. The experimental setups are straightforward and compact. CGH has gotten a lot of attention from academia and industry. It is the future form of 3D display.

In theory, computer-generated holography converts a three-dimensional (3D) object into a digital two-dimensional (2D) hologram using diffractive calculations. The 2D hologram is then uploaded to a plane wave-illuminated spatial light modulator (SLM). At a certain distance, the optical reconstruction of the 3D object is obtained.

Computer-generated holography could be used in various 3D displays, including head-mounted, heads-up, and projection displays. How to generate high-speed and high-quality 2D holograms is a critical issue and research direction in this field.

Researchers recently proposed 4K-DMDNet, a model-driven deep learning neural network. It achieves high-fidelity 4K color computer-generated holography displays and high-quality, high-speed hologram generation. Because of SLM’s limitations, the calculated complex-amplitude distributions on the holographic plane must be converted into amplitude-only or phase-only holograms (POHs).

Read more

Related Content: Holographic Algorithms – Drawing Lines In 3D Space