Find out the peak position from graph

[M.Ali]

Hello Klemen,,,

Firstly: appreciate your support,,,

So I am keeping try to use the suggested algorithm,,, but it will lead to the same problem,, that means my algorithm and your's had limitation for some extent in my application,,,,

i will  see the posibility of find out the positions without using image processing ( directly from the graph),,, because image processing depends on gray level which is differen for different images and that implies different threshold value....

thank you again for your time,,,also thank you Vekkuli,,

[BruceAmmons]

Here's a totally different idea:

Start with the original image.  Get the average intensities in the X and Y axis.  For each axis, use correlation to figure out the shift.  There should be a distinct repeating pattern.  If you can find the shift in the X and Y axis, simple math gives you the diagonal.  This would probably give you the same result as your 2D correlation, but much faster and easier.

Bruce

[BruceAmmons]

Another idea:

Step through the image in 25x25 blocks (or whatever size makes sense).  They could overlap if desired.  Evaluate each block to see if it makes a good template for correlation (good contrast, etc.).  If it does, use it for correlation on the entire image.  You should get two peaks if it is part of a particle.  Measure the distance.  Repeat for the entire image.  Use statistics to get the correct distance.  Perhaps a weighted average of distance based on peak intensity.  Or median distance.

This would eliminate the huge peak in the middle.  The two other peaks could be found easily by projecting to the X and Y axis and using peak finder.

Bruce

[M.Ali]

Hello Bruce,,,, i can't understand well your ideas,, there is some thing missing for me but i don't know what is it exactly,,, in the  first idea : to apply correlation function we shoud subtract the mean value from the raw image because it doesn't include information,,,so i think you mean by average intensity the mean intensity,, if that is right ,i do that in my algorithm,,, So  please can you explain that 2 iseas as simple block diagrams,, i will be grateful,,,,

thanks