Star removing neural net software

For a variety of reasons, it is often desirable to remove stars from an astronomical image. This could be so that any background nebulosity can be stretched a little more without saturating the stars and losing their colours. The stretched image can then be combined with the original image or some other procedure, to put the stars back into the image. Other reasons might be so that the structure of a nebula can be studied without the distraction of the stars, or for example, to study the structure of a comet without the interference of the background stars.

Here I have used Starnet++, a neural net that has been trained to remove stars from images taken through refractors. More details on what is happening is given HERE.

Here are two animations alternating between the original image and the image with stars removed by Starnet++.

Click on an image to get a closer view.

The Orion Nebula

Comet C/2020 F3 (NEOWISE)

It is interesting to see how this aspect of AI is useful for astronomical image processing.

Steve Wainwright