基于复数神经网络的单偏振调制的偏振复用多模光纤成像系统
By employing a single-polarization modulation method based on complex neural networks, the problem of polarization mode coupling in multimode fiber imaging is solved, achieving polarization multiplexing and efficient image projection, simplifying the optical system and reducing its complexity.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- FUDAN UNIVERSITY
- Filing Date
- 2023-06-07
- Publication Date
- 2026-07-17
AI Technical Summary
In existing multimode fiber imaging technology, the polarization mode coupling problem after the light field is transmitted through multimode fiber has not been effectively solved, resulting in the loss of polarization characteristics. Existing neural network models cannot reasonably model the complex characteristics of the light field and the polarization mode coupling phenomenon.
A single polarization modulation method based on complex neural networks is adopted, and a complex projection network (CV-Pnet) is designed. Polarization multiplexing is realized on a polarization-sensitive spatial light modulator through a generative-model neural network structure. By utilizing complex fully connected layers and angle-taking operations, the algorithm space complexity is reduced and the optical path system is simplified.
It achieves the preservation of polarization characteristics after transmission through multimode fiber, improves the confidence and convergence speed of image projection, reduces the risk of overfitting, and simplifies the structure of the optical system.
Smart Images
Figure CN116782042B_ABST