Find notes into a stave

Hello everyone. I need some help with a program which consist in detect and extract notes into a stave ONE BY ONE. All the notes are simply notes (Quarter Note,Half Note,Whole Note,Eighth Note,Sixteenth Note). I need to detect the possition x,y, and count every note in the stave.I only need to know the position of the head of every note... The stave is a controlled stave, with the same dimmension and the position of the lines are always the same, no rotation, no distorsion... Could anyone help me please, or give me some ideas?

Answer 1

[Matt J]

I would try to remove the staff, leaving only the notes behind. Then I would analyze their shape and location with regionprops(). For example, below is a crude binary image of a staff with a quarter note. I can produce an image B with the staff lines removed by looking at the local thickness of the black regions and checking if that thickness is more than the thickness of a staff line,

   lthick=conv2(A,[1;1;1],'same'); %local thicknesses
   mask=imdilate((lthick==1)&A,[1;1;1]);

   B=A&~(mask),

A =

       1     1     1     1     1     1     1     1     1     1
       0     0     0     0     0     1     0     0     0     0
       0     0     0     0     0     1     0     0     0     0
       0     0     0     0     0     1     0     0     0     0
       1     1     1     1     1     1     1     1     1     1
       0     0     0     0     0     1     0     0     0     0
       0     0     0     0     0     1     0     0     0     0
       0     0     0     1     1     1     0     0     0     0
       1     1     1     1     1     1     1     1     1     1
       0     0     0     1     1     1     0     0     0     0
       0     0     0     0     0     0     0     0     0     0
       0     0     0     0     0     0     0     0     0     0
       1     1     1     1     1     1     1     1     1     1
       0     0     0     0     0     0     0     0     0     0
       0     0     0     0     0     0     0     0     0     0
       0     0     0     0     0     0     0     0     0     0
       1     1     1     1     1     1     1     1     1     1

[Alberto]

Thanks for the idea, I will try with this line. Currently I am using a mask (the note, for example black note) which compare in whole the image looking for a coincidence. The problem is that the image came from other transformation like rotation, scale etc.. and it is so deformed that the mask does not match...

Answer 2

[Image Analyst]

People have written papers on this so I'm not going to spend time redeveloping an existing algorithm for you. Go here http://iris.usc.edu/Vision-Notes/bibliography/other1021.html#Music to find successful algorithms that people have developed and published and are using.

[Alberto]

Excuse me if I made you to waste your time... I think that rude modals were not necessary... I have used Matlab for 2 months only... I can´t know everything. Anyway, thank you very much for your answer.

[Image Analyst]

Alberto, if you visit here much, you'll see that I very very frequently develop custom algorithms for people and post the code. But in those cases, it's usually for something very simple that I don't have to spend more than a few minutes developing, or for demos that I already have saved on hand. Yours did not look like it fell into those categories so I wanted to point you to some places where you could find your answer, rather than wait for me to do it (like so many do). Whatever I would do in a few minutes would undoubtedly not be as good as the published methods. I thought you might like a good robust method rather than some prototype small snippet of code I hacked out in a few minutes so I thought I'd let you know where you can find them. Many people are not aware of this valuable database where virtually all image processing articles are archived . Good luck with it.