Wavefront-coded image restoration technology based on multi-scale deep neural networks

Area of Science:

• Optics and Photonics
• Artificial Intelligence
• Aerospace Engineering

Background:

• Aerodynamic thermal effects significantly degrade optical imaging system performance during flight.
• Image blurring caused by these effects poses a challenge for accurate visual data acquisition.
• Existing restoration methods may struggle with the dynamic and complex nature of flight-induced aberrations.

Purpose of the Study:

• To develop a novel wavefront-coded image restoration method for optical systems.
• To address image blurring caused by aerodynamic thermal effects under varying flight conditions.
• To improve the quality and detail recovery in restored optical images.

Main Methods:

• Proposed a multi-scale deep autoencoder neural network (MS-DAE) for wavefront-coded image restoration.
• Implemented a multi-scale loss function with residual attention mechanisms.
• Modulated blur levels to train and evaluate the network's performance.

Main Results:

• Achieved significant improvements in Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index Measure (SSIM) compared to Wiener filtering.
• Demonstrated superior performance over the BaseNet model in PSNR and SSIM.
• Successfully restored image details and suppressed artifacts in reconstructed images.

Conclusions:

• The proposed MS-DAE method effectively restores images degraded by aerodynamic thermal effects.
• The approach shows adaptability to diverse flight conditions, indicating practical applicability.
• This technique offers a promising solution for enhancing optical imaging in aerospace applications.