Compensation of Atmospheric Turbulence Phase Distortion Using Neural Network Adaptive Optics

Abstract

Adaptive optics revolutionizes telescopic resolution but faces cost, complexity, and calibration hurdles. Neural network adaptive optics (NNAO) offers promise by using neural networks to tailor corrections to telescopes and atmospheric conditions, by passing calibration and sensors. This MATLAB-based study examines NNAO's impact on astronomical image quality, revealing it as a cost-efficient solution that enhances adaptive optics in astronomy. The numerical simulation results were encouraging, with a compensation rate exceeding 50% due to favorable monitoring conditions. The results indicate that the dominant factor affecting image quality is the variance of wavefront aberrations. The Strehl ratio (SR) decreases from an average of 0.548 for a variance of 0.2 to 0.020 for a variance of 0.6, while the mean squared error (MSE) increases from an average of 0.613 to 5.414. However, the effect on peak signal-to-noise ratio (PSNR) is inconclusive. Furthermore, it was found that increasing the number of neurons and training ratio does not significantly impact the results obtained, but it noticeably affects the computational time required.