What is stellar interferometry?
Stellar interferometry is an observational technique that makes use of more telescopes to obtain a picture of a star or a stellar object.
A stellar interferometer measures the coherence degree between couples of telescopes (single baselines) observing a distant object like a star. The correlator then gets those measures (called “visibilities“) that must be placed on a perspective plane depending on their position and inclination (perspective distance of the telescopes given the inclination of the star). Once this plane is filled enough with the visibilities an inverse fourier transform must be calculated from this bidimensional picture to obtain an aproximation of the shape of that star.
This technique, called “aperture synthesis” (synthesis of an aperture) does a calculation of the perspective distance between the telescopes, adjusts the light arrival time delays between them, and fills the perspective plane with the visibilities taken. Later an inverse Fourier transform can be perfomed to this plane to synthesize an image of the object observed.
The telescopes can be placed without being moved, in this way the perspective distance between the telescopes changes with time due to the Earth rotation. This method is widely used in astronomical stellar interferometers. Since the rotation of earth takes about 24 hours to complete a 360 degrees rotation, the time needed to fill the perspective plane with the visibilities is often very long.
What is coherence degree and why it is useful?
I wrote here about the coherence degree. What is it? Coherence degree is the correlation between two fluxes or elements taken for comparison. In stellar interferometry a high coherence degree between two fields of view means that the fields observed have many things in common. Intensity interferometry, for example, measures the intensities only, so the mean luminosity of an observed area is compared with the mean luminosity of another one.
More advanced intensity interferometers compare or count the arrival time of intensity fluctuations and count the synchronously measured peaks read out as voltage pulses.
How images are obtained?
The visibility plane then shows the coherence ratios only, so the similarities at various baselines lengths. This means that there are no direct pictures shown in these plots (maps). Only a count of symmetries in the space domain. The answer is to try an inverse Fourier transform of the map, so to obtain the raw picture that generated these coherence which depend on the place where they were taken. These maps, however, are not always fully plotted. They need some models that complete them.
What are models and how to use them?
Since most of the maps (Fourier planes) are still incomplete after many observations also, models are used to complete them. A good model is a complete map of the Fourier transform of the object observed picture.
Models should be mixed to the observation for estimation of the complete picture and aimed to obtain a nice and realistic inverse transform.
They can also be generated by an estimation of the missing parts of the map, and AI and machine learning are used to improve the results in a more realistic manner here.