Crosstalk-free for multi-plane holographic display using double-constraint stochastic gradient descent

Area of Science:

• Optics and Photonics
• Computer Vision
• Computational Imaging

Background:

• Multi-plane crosstalk degrades holographic 3D display quality.
• Existing time-multiplexing stochastic gradient descent (TM-SGD) methods struggle with decreased inter-plane intervals and increased plane numbers.
• High crosstalk and long optimization times limit applications like tomographic 3D visualization.

Purpose of the Study:

• To propose a novel double-constraint stochastic gradient descent method for suppressing inter-plane crosstalk in multi-plane holographic reconstruction.
• To enhance reconstruction quality and reduce computational time compared to existing methods.

Main Methods:

• Implemented a double-constraint stochastic gradient descent approach.
• Utilized a mask to focus optimization on signal regions.
• Applied phase regularization to reduce signal region phase randomness and suppress crosstalk.

Main Results:

• The proposed method effectively suppresses inter-plane crosstalk, especially at lower inter-plane intervals.
• Reconstruction quality is significantly improved.
• Optimization time is reduced by approximately 4 times compared to TM-SGD.
• Numerical simulations and optical experiments validate the findings.

Conclusions:

• The double-constraint stochastic gradient descent method offers superior performance in multi-plane holographic reconstruction.
• This advancement is crucial for high-quality tomographic 3D visualization in biomedical fields.
• The method enables efficient and accurate reconstruction of multiple tomographic images without crosstalk.