Researchers from Chiba University in Japan are using artificial intelligence (AI) deep learning to holographic 3D images taken directly from regular 2D color images using standard color cameras.
The method employs three deep neural networks (DNNs) to transform regular 2D color images into data that then can be displayed as a 3D scene or object as a hologram.
The first DNN uses the color image captured using a regular camera as the input and predicts the associated depth map. The original RGB image and the depth map created by the first DNN are then used by the second DNN to generate a hologram. The third DNN refines the hologram generated by the second DNN to make it suitable to be displayed on different devices.
Previous methods have been proposed for generating holograms directly from 3D data capture using RGB-D cameras that capture both color and depth information. But no method has previously been proposed using standard 2D images and then create a hologram.
“There are several problems in realizing holographic displays, including the acquisition of 3D data, the computational cost of holograms, and the transformation of hologram images to match the characteristics of a holographic display device,” said Tomoyoshi Shimobaba, professor at the Graduate School of Engineering at Chiba University. “We undertook this study because we believe that deep learning has developed rapidly in recent years and has the potential to solve these problems.”
What are holograms?
Holograms provide a level of detail that is unavailable from regular 2D images. They offer realistic views of 3D objects more so than regular 2D images.
Traditionally, holograms are constructed using three-dimensional data of an object and the interactions of light with an object. The problem is this method is computationally intensive and requires the use of a special camera to capture the 3D images. This presents a challenge for the generation of holograms and limiting their widespread use.
The Chiba University team found that the approach to process data and generate the hologram was superior to using state-of-the-art GPUs.
Additionally, Shimobaba said the reproduced image of the final hologram can represent a natural 3D reproduced image, which is inexpensive and does not require an imaging device like an RGB-D camera.
Potential use cases
Researchers believe holograms could be potentially used in fields like medical imaging, manufacturing or virtual reality.
Holograms could also find potential in automotive applications such as heads-up and head-mounted displays that generate high-fidelity 3D displays. It could even potentially generate in-vehicle holographic images for drivers to view people, roads and signs in 3D.
The full research can be found in the journal ScienceDirect.
