Chinese Optics Letters, Volume. 23, Issue 10, 100501(2025)
Optimized binary computer holography via convolutional neural network-based differentiable binarization
Fig. 1. DMD holographic near-eye display scheme based on double-step band-limit diffraction.
Fig. 2. (a) Flowchart of the SGD optimization algorithm for generating binary CGH. (b1), (b2) The binarization network architecture.
Fig. 3. (a) Target images. (b)–(e) Simulated comparison of different binary CGH methods, including (b) direct binarization to SGD optimized amplitude hologram, (c) Func-B-SGD optimization with clip mapping binary-similar function, (d) Func-B-SGD optimization with curve mapping binary-similar function, and (e) our proposed CNN-B-SGD optimization.
Fig. 4. (a) Optical setup of DMD holographic near-eye display. (b) Comparison of experimental results of different methods. (c) Experimental results of holographic display using our CNN-B-SGD-based binary CGHs for the 2D image case. (d) Experimental results of holographic display using our CNN-B-SGD-based binary CGHs for a true 3D case when the camera is focusing from 400 to 700 mm. Intensity image and corresponding depth map of the 3D scene are illustrated.
Fig. 5. Experimental results of full-color holographic display using our CNN-B-SGD based binary CGHs.
|
Get Citation
Copy Citation Text
Jiadi Shi, Shuqing Cao, Xian Ding, Bo Dai, Qi Wang, Songlin Zhuang, Dawei Zhang, Chenliang Chang, "Optimized binary computer holography via convolutional neural network-based differentiable binarization," Chin. Opt. Lett. 23, 100501 (2025)
Category: Diffraction, Gratings, and Holography
Received: Apr. 11, 2025
Accepted: Jun. 10, 2025
Published Online: Sep. 17, 2025
The Author Email: Chenliang Chang (changchenliang@hotmail.com)