Imaging Through Turbulence: A Computational Approach

The night sky is not viewed through a uniform optical medium, but through a constantly shifting mosaic of air pockets with varying temperature, pressure, and density. These fluctuations alter the refractive index of the atmosphere, distorting incoming wavefronts and degrading the spatial resolution of ground-based imaging systems. Traditional adaptive optics (AO) systems mitigate these effects in real time using a wavefront sensor, control system, and deformable mirror. However, AO performance is limited by actuator density, calibration requirements, the need for guide stars, reduced effectiveness at low elevation angles, and across wide fields of view.

This project presents preliminary work into a computational approach to atmospheric turbulence correction that operates independently of deformable optics. Rather than treating turbulence purely as a stochastic phase error, this method analyzes its temporal evolution by integrating both 'image' and frequency spatial information. Using this dual-domain sampling framework, we track how localized refractive index variations induce phase errors in the spatial frequency components over time and computationally correct these phase errors before image reconstruction.

Our laboratory experiments generate controlled turbulence over a 5-meter air path using a laser beam. Beam wander at the detector plane is analyzed as a function of time, revealing that centroid motion is primarily driven by a series of field-scale phase tilts induced by turbulence along the beam path. Longer pathlength data was collected using a portable telescope under real night-sky conditions. These observations are used to inform simulations that replicate the apparent motion and distortion of celestial targets, with the goal of computing out distortion and image 'jitter' caused by turbulence.

An adaptive optics setup was used in parallel laboratory tests to benchmark performance. This method may serve as a standalone partial-correction technique or to complement existing AO systems, particularly in cases where traditional hardware struggles.