Instability in pattern matching

[muks]

Can you post a few images?

[FM82]

I'm trying to get a fail so i can post the "good" and the "bad".

[muks]

---

@FM82 wrote:

I'm trying to get a fail so i can post the "good" and the "bad".

---

Also post the fiducials as the camera sees it.

[FM82]

Here it is. i'm sorry for the patch but i need to hide the client's name. Here you can see the fiducials and the areas of search with the relative components. This is a good image. After the modification that i explained before i had done 100 runs but i could not get a single "fail", instead i have obtained a pretty costant match score.

[muks]

and what is your reference in this image?

[FM82]

Sorry i did'nt get it, do you mean the fiducials?

[muks]

yes

[FM82]

The three  points that you can find in the upper left corner, in the lower left and in the upper right.

This one:

[muks]

I honestly feel is the image is good. I feel if you are taking the fiduals reference corerctly the application should work correctly and ofcource with uniform lighting.

[FM82]

I already checked the fiducials and i can exclude that they are the problem. Honestly last week there had been some cloudy days with the sun that come and gone from time to time. So i think that the principal problem is my inexperience with the pattern matching function and the working mechanism behind the result, so i expect something that i couldn't really achieve due the terrible conditions that i experienced last week.
Today is a really sunny day and it seem that the code is working pretty well. I work really close to a window so i take all the changing in lighting. I know that this are terrible conditions to work with, when talking about image analysis, but i thought that if i can make a software that can work under this circumstances when i will put all the stuff inside a closed dome i will be sure that everything will work very well.