Turbulence Mitigation in Optical Imaging Using Pyramid Attention Gan

Abstract

Atmospheric turbulence, a common issue in optical transmission, introduces significant blurring and geometric distortion, which degrade image quality and complicate subsequent analysis. This challenge is particularly pronounced in applications such as satellite imaging, security surveillance, and aerial photography, where accurate image restoration is crucial. To address these issues, we propose a novel deep learning framework, the Pyramid Attention Generative Adversarial Network (PA-GAN). This approach integrates a pyramid structure with an attention mechanism to capture multi-scale features and improve feature extraction precision. The Generative Adversarial Network (GAN) component further refines the imaging process by generating high-quality outputs through adversarial training. Experimental results demonstrate that PA-GAN outperforms existing methods in terms of both Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index (SSIM), significantly reducing blurring and distortion, and enhancing image clarity. The proposed method not only provides an effective solution to turbulence mitigation but also represents a significant advancement in deep learning-based image restoration techniques.

Keywords: atmospheric turbulence, Optical Imaging Restoration, deep learning, Generative Adversarial Networks, Pyramid Attention Mechanism