Reconstruction of distorted underwater images using robust registration

Area of Science:

• Optics and Photonics
• Computer Vision
• Image Processing

Background:

• Imaging through dynamic air-water interfaces presents significant challenges due to unpredictable light ray refraction.
• These refractive effects cause complex geometric distortions, hindering image quality and analysis.

Purpose of the Study:

• To develop and present a novel algorithm for undistorting dynamic refractive effects in images captured through fluctuating air-water surfaces.
• To enhance the quality of images affected by wave turbulence and refractive distortions.

Main Methods:

• An iterative robust registration algorithm is utilized to align each frame to a reference frame, compensating for wave-induced turbulence.
• A high-quality reference frame is reconstructed using selected patches from the image sequence.
• A blind deconvolution algorithm is applied to further refine the reference frame quality.

Main Results:

• The proposed algorithm effectively corrects geometric distortions caused by dynamic refractive effects.
• Experimental results demonstrate significant improvements in image quality compared to existing methods.
• The reconstructed reference frame provides a stable and clear basis for distortion correction.

Conclusions:

• The developed algorithm offers a robust solution for imaging through turbulent air-water surfaces.
• This technique has the potential to improve various applications requiring clear imaging in challenging aquatic or surface environments.
• The method shows superior performance in mitigating dynamic refractive distortions.