一种基于生成对抗网络的散斑图像还原方法及系统
By employing a generative adversarial network architecture and a multi-objective optimization strategy, combined with residual blocks and cross-scale feature fusion, the problem of speckle noise removal in multimode fiber imaging was solved, achieving high-quality image restoration and reconstruction.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- TIANJIN UNIVERSITY OF TECHNOLOGY
- Filing Date
- 2025-06-16
- Publication Date
- 2026-07-17
AI Technical Summary
Existing technologies struggle to effectively remove speckle noise in multimode fiber imaging, leading to severe distortion or loss of image information and limiting the application of high-quality imaging.
A generative adversarial network architecture is adopted, which combines a generator with residual blocks and cross-scale feature fusion, a PatchGAN discriminator and spectral normalization mechanism, and performs image restoration through a multi-objective optimization strategy. Large receptive field convolution and sub-pixel convolution layers are used for feature extraction and reconstruction.
It significantly improves the ability to restore image details and the network generalization ability, and realizes high-resolution, clear image reconstruction, which is suitable for high-quality image reconstruction tasks in multimode fiber imaging systems.
Smart Images
Figure CN120339136B_ABST