Speckle imaging

Speckle imaging is a set of high-resolution astronomical techniques that work by taking many very short exposures, which freeze the random motion of atmospheric turbulence. These methods can greatly improve the resolution of ground-based telescopes, but they only work well on bright targets. The two main categories are the shift-and-add (or image stacking) method and speckle interferometry. The core idea behind all these techniques is to capture extremely brief images of celestial objects and then process them to cancel out the blurring effects of the atmosphere.

This approach has led to many discoveries, such as thousands of binary star systems that would otherwise look like a single star through a similar-sized telescope, and the first images of sunspot-like phenomena on other stars. Many of these methods are still widely used, especially for imaging relatively bright objects. A telescope’s resolution is normally limited by the size of its main mirror due to Fraunhofer diffraction, which spreads a point of light into a small spot called the Airy disk.

Objects closer together than this spot’s size appear as one. Larger mirrors can therefore resolve finer details, but the atmosphere disrupts this: instead of a single Airy disk, it creates a pattern of similarly-sized spots spread over a larger area. Under good visible-light conditions, the effective resolution limit is set by the atmospheric seeing parameter r0—about 20 centimeters in mirror diameter. For years, this limited telescope performance until speckle interferometry and adaptive optics came along.

Speckle imaging reconstructs the original image through processing. The key insight, from American astronomer David L. Fried in 1966, was to take very fast images that freeze the atmosphere in place. At infrared wavelengths, the coherence time τ0 is around 100 milliseconds, but in visible light it drops to as little as 10 milliseconds. If exposures are shorter than τ0, the atmosphere is too slow to blur the image, and the recorded speckles are a snapshot of the seeing at that instant.

Coherence time τ0 = r0/v depends on wavelength because r0 does. The downside is that such short exposures are hard to capture, and if the target is too dim, not enough light reaches the detector for analysis.