Lizzy’s Status 3/2

Lizzy’s Status 3/2

This week I was working on finding the note heads on each component. I already found all the connected components last week, and now, while there are more than just note components, those are the vast majority of the important things I need to identify. So, being able to find note heads is vital, especially since there can be more than one per component and knowing where they are as well as how many there are determines a large amount about the music.

After reading through several of the papers again, one of the suggested methods of finding the note heads was Hough Circle Transform. This is an algorithm that basically allows you find circle-ish shapes in an image based on several parameters including the min and maximum circle radii you are looking for, how sharp the edge needs to be, how close to a circle it needs to be, etc.

After playing around with it for a long time, I found the openCV method for hough circle transform to be about right with the right parameters. It took a decent amount of fine tuning, but I am able to reliably calculate where note heads are in each of the components. As I look at more complicated pieces of music, these again might need to be adjusted some, but for now it seems reasonable.

Now that I have noteheads, I would like to be able to determine what type of note each one is as well as pitch. Pitch should be fairly easily attainable by using the center y values for each of the components to determine where on the staff it falls. Note type may be more difficult, especially when I need to find eighth vs sixteenth, etc. and they are all connected together.

If I have time (hopefully I will next week), I will also have some time to further fine tune the parameters for the circle transform, as well as look into template matching for certain types of notes like rests, clef symbols, etc. that don’t have circles involved in them, but have very distinct shapes.

I am also still on time, so nothing about the schedule truly needs to be changed or adapted. Our group has decided to be primarily finished with each of our major parts by the end of spring break so that we can work on putting the pieces together for the mid-semester demo.

Leave a Reply

Your email address will not be published. Required fields are marked *