<< Chapter < Page Chapter >> Page >

Matched-filtering

Our program was tailored to work flawlessly on our self-generated sheet music. Under our provided boundary conditions, all music notes were correctly identified and converted into the data parameters of length, octave, and frequency.
The first picture is a self-generated unprocessed sheet of music that we intend to analyze:
mixed
The second picture shows the correlation gradient after applying a whole notes matched filter. The red X's indicate where the hits are:

xcorr

Finally, the third picture shows the hits back on the original sheet of music:

detected1

We also tried our program on a sheet of sheet music that we scan into the computer. Before running our program on the image we made sure the staffs were oriented correctly and that the image was close to the same scale as our self-generated pictures. The following picture shows that our program shows hits for quarter notes accurately:

oldmc

As a note to the reader, our program has trouble recognizing eighth-note tails, but they were found when the tolerance level was lowered.

Get Jobilize Job Search Mobile App in your pocket Now!

Get it on Google Play Download on the App Store Now




Source:  OpenStax, Optical character recognition. OpenStax CNX. Apr 15, 2011 Download for free at http://cnx.org/content/col11296/1.1
Google Play and the Google Play logo are trademarks of Google Inc.

Notification Switch

Would you like to follow the 'Optical character recognition' conversation and receive update notifications?

Ask