Shulou(Shulou.com)11/24 Report--

Thanks to CTOnews.com netizen Alejandro86 for the clue delivery! CTOnews.com, October 22 (Xinhua), holograms can provide three-dimensional (3D) visual effects of objects, which are more realistic and three-dimensional than two-dimensional (2D) images, and are of great value in medical imaging, manufacturing, virtual reality and many other fields.

CTOnews.com noted that traditional holographic techniques need to record three-dimensional data of objects and interact with light, which requires high computing power and special cameras to capture three-dimensional images, which limits the wide application of holograms.

In recent years, many methods based on deep learning have been proposed to generate holograms. They can create holograms directly from 3D data captured by RGB-D cameras, and RGB-D cameras can obtain color and depth information of objects at the same time. This method avoids many computational challenges in traditional methods, and is a simpler method to generate holograms.

Now, a research team led by Professor Yoshiyuki Hashimoto of the Institute of Engineering at Chiba University in Japan has proposed a new method based on deep learning, which further simplifies the process of generating holographic images. three-dimensional scenes or objects can be generated directly from two-dimensional color images captured by ordinary cameras. The study was recently published in the journal Optical and Laser Engineering (Optics and Lasers in Engineering). Kaisuke Ishii and Tomoshi Ito of the Institute of Engineering at Chiba University were also involved in the study.

Speaking about the motivation behind this research, Professor Xiagu said: "there are several difficulties in realizing holographic display, including the acquisition of three-dimensional data, the computational cost of holograms, and the conversion between holographic images and holographic display device characteristics. We conducted this research because we believe that deep learning has the potential to solve these problems in recent years."

In this method, three depth neural networks (DNN) are used to convert ordinary two-dimensional color images into holograms that can be used to display three-dimensional scenes or objects. The first DNN uses color images captured by ordinary cameras as input, and then predicts the relevant depth maps to provide information about the three-dimensional structure of the image. The original RGB image and depth map generated by the first DNN are used by the second DNN to generate the hologram. Finally, the third DNN optimizes the hologram generated by the second DNN to make it suitable for display on different devices.

The researchers found that this method takes more time to process data and generate holograms than state-of-the-art graphics processing units. " Another advantage of our method is that the images reproduced by the final hologram can show a natural and real three-dimensional effect. Moreover, depth information is not needed to generate holograms, so this method is low-cost and does not require the use of 3D imaging devices such as RGB-D cameras. " Professor Xiashi added as he further discussed the results.

In the near future, this method can find potential applications in head-mounted displays and help to generate high-fidelity 3D displays. Similarly, it is possible to innovate the vehicle holographic display to display necessary information such as pedestrians, roads and signs in a three-dimensional way, which is expected to pave the way for promoting the popularization and development of holographic technology.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.