Fast super-resolution with affine motion using an adaptive Wiener filter and its application to airborne imaging

Area of Science:

• Image processing
• Computer vision
• Signal processing

Background:

• Traditional super-resolution (SR) methods struggle with non-translational motion due to assumptions about warping and blurring.
• Airborne imaging often involves complex motion beyond simple translation, limiting existing SR techniques.

Purpose of the Study:

• To develop a novel SR algorithm capable of handling non-translational motion, specifically affine motion, for applications like airborne imaging.
• To address the commutation assumption limitations in existing SR methods.

Main Methods:

• Fourier domain analysis to demonstrate approximate commutation between affine warping and the point spread function.
• Development of a fast adaptive Wiener filter (AWF) SR algorithm incorporating a smart observation window.
• Precomputation of filter weights for efficient processing of various motion types.

Main Results:

• The proposed Fourier analysis shows approximate commutation for affine motion, validating the new approach.
• The fast AWF SR algorithm demonstrates effective performance with affine motion.
• Evaluation using simulated and real infrared airborne imagery, including a thermal resolution target, confirms objective resolution analysis.

Conclusions:

• The developed fast AWF SR algorithm successfully extends super-resolution capabilities to non-translational motion, particularly affine motion.
• This advancement is crucial for improving image quality in applications like airborne surveillance and remote sensing.