Hybrid deep learning framework for image restoration in Fso systems affected by log-normal fading

Area of Science:

• Optical Engineering
• Image Processing
• Artificial Intelligence

Background:

• Atmospheric turbulence causes log-normal fading, degrading image quality in Free Space Optical (FSO) systems using high-order modulation like 64-QAM.
• Traditional methods like MIMO and adaptive optics offer limited restoration effectiveness in dynamic environments.

Purpose of the Study:

• To develop a novel hybrid deep learning framework for robust image restoration in FSO communication systems.
• To address the limitations of conventional techniques in mitigating atmospheric fading and noise.

Main Methods:

• A hybrid framework combining a Deep Convolutional Neural Network (DCNN) for denoising and image sharpening for clarity enhancement.
• Training the DCNN on a synthetic dataset of 30,000 images with varying Signal-to-Noise Ratio (SNR) levels to learn log-normal fading suppression.

Main Results:

• Achieved substantial improvements in Peak Signal-to-Noise Ratio (PSNR), increasing from 19 dB to 68 dB.
• Outperformed conventional methods in both quantitative (PSNR, RMSE) and qualitative assessments.
• Successfully restored structural image content with high fidelity.

Conclusions:

• The proposed hybrid deep learning approach effectively restores images degraded by atmospheric distortion in FSO systems.
• Demonstrates high efficacy for real-world applications including medical imaging, remote sensing, and surveillance.