Hot Pixels

Image processing, astrometry, photometry, etc.
Post Reply
Forum_2016
Posts: 219
Joined: 08 Dec 2018, 13:30

Hot Pixels

Post by Forum_2016 »

Hello,

Is it possible to explain:

-how pixels are determined as hot, or cold, or a group?

-how these hot or cold pixels, or groups are then treated?

Regards,

Patrick

Forum_2016
Posts: 219
Joined: 08 Dec 2018, 13:30

Re: Hot Pixels

Post by Forum_2016 »

Hello,
they are determined by statistical analysis, compared to their neighbours.

To be considered "single" they must be the only pixel in a 3x3 square. When this is not true, for example two hot pixels in a row, they are not corrected, unless "group" is selected. This can delete/affect some stars though.
The correction is very easy, the hot pixel is replaced by the average value of its neighbours.

For DSLRs and color CCDs that is not true, the filter need to compare the values with the "same color" (R,G,B) which are usually inside a 5x5 square around the candidate pixel.

Kind regards,
Fabio.

Forum_2016
Posts: 219
Joined: 08 Dec 2018, 13:30

Re: Hot Pixels

Post by Forum_2016 »

Fabio,

Thank you.

Are you able to provide some descriptive details of the statistical analysis for "singles"?

Regards,

Patrick.

Forum_2016
Posts: 219
Joined: 08 Dec 2018, 13:30

Re: Hot Pixels

Post by Forum_2016 »

Hi, it's very similar to Sigma, the neighbour pixel are considered, then if one value is really far from the standard deviation of the group, it's detected as an hot pixel. You can see the algorithm at work if you enable the real time preview and move the slider (which controls the standard deviation). The exact algorithm is a small secret but you can find many examples with an internet search. The algorithm for "groups" is instead a modified median filter.

Forum_2016
Posts: 219
Joined: 08 Dec 2018, 13:30

Re: Hot Pixels

Post by Forum_2016 »

Fabio,
Thank you.
Patrick.

Post Reply