Jump to content


Using AI for imaging


Recommended Posts

Hi People,

Has anyone used machine learning to help with imaging, eg guiding and PEC or aligning and stacking, picking out faint images etc?

Also I'm thinking focusing, drizzle and dithering.

What are people's thoughts?



Edited by SteveBz
Link to comment
Share on other sites

Define what you mean by AI ??  its' a very hip\misused\understood term for manipulation of computer data (of any source\type) to try & enforce a required output, so all of us that use guiding, plate solving, imaging etc. uses AI

Link to comment
Share on other sites

Well I'm using open CV feature detector to help me to focus. It automatically chooses a star and then I use ordinary python code to adjust the focus, feed it back into open CV again and open CV gives me a HFD-like update.

I thought that some sort of machine learning of the sort that Google do with their stuff like tensorFlow might speed it up and avoid me having to try algorithm after algorithm to find the optimal case.

Link to comment
Share on other sites

So from your example, if you recorded the exact focus position, each time you go through the process, and used that data as a starting point for the routine, then you would be using a basic form of AI...

Plate solving works on the basis of a database of stars, combined with a knowledge of where the mount needs to be pointed from its' starting point, and iteritivly takes an image of where the mount is pointing and compares this image with the database, and adjusts the mount to match the database and local image....

All you would be doing by trying to include a large corporate in the process, would be to give them the ability to target you with appropriate advertising on your obvious preferences.... no thanks....



Link to comment
Share on other sites

There are some positive results with regards to image enhancement already.


It won't be much longer before the same type of convelutional networks can generate higher resolution results. A few recent papers came out about dynamically growing networks during training, rather than starting with a high res image from the start.

Beyond image enhancement, I suspect that we will see someone come up with a full image guiding approach. Using some kind of recurrent neural network to determine which direction your mount needs to move in to stack locked on a target.

  • Like 1
Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue. By using this site, you agree to our Terms of Use.