After the findstar, two sets of star point lists are obtained. Why did the earlier version use all the star points from both sets for matching, while the later version chose to use the smaller set of star points for matching?
Hi,
you will have to be a bit more specific… which portion of the code are you refering to?
C.
Hi， In matching the stars after the peaker, I noticed that the new_star_match section in the source code of the siril-master version differs from that in the siril-1.2.3 version. In the siril-master version, the list of stars is extracted entirely from both groups for subsequent transformation calculations, while in the siril-1.2.3 version, only the smaller group of stars is extracted, and the larger group is limited to extracting the same number of stars as the smaller group for the transformation calculations. I would like to know the reasoning behind this approach.
this is siril-master version：
this is
siril-1.2.3 version
thanks！
ah ok,
so in master, we are using the full list on each image so as not to reject possible legit matching stars further down the process.
This has to be read in conjunction with the changes made in atFindTrans. Say A is the short list, and B the long list. We are making more triangles for B to catch more votes for stars which may be further down in B but actually be matching stars in A. If we were to cut B to the same number as A, we may have bright stars in B not present in A that take the place for dimmer stars in B that have a match in A.
C.
It’s interesting to see someone so deeply interested in this part of the code, which isn’t the easiest to understand. How come you’re so interested?
HI，recently, I have been learning about astronomical image alignment, so I am more interested in this piece.
Thanks.I will pay attention to the updated content of atfindtrans later.