The study and prototype system of printed music recognition

Abstract

Optical Music Recognition (OMR) is the automatic recognition of a scanned page of printed music notation by computer. A new approach to this pattern recognition problem is presented in this paper. The concept of direction number has been defined according to the musical notation's statistic and structural characteristic, and then make use of mathematical morphology theory to process musical score image and recognize musical information. Specialized algorithms are used firstly to preprocess a musical score image and then recognize stafflines, barlines, pitch, note values, clef, etc.. Finally the musical information of the musical score image is stored in the MIDI format. Making use of our proposed algorithms, which can process multiple staves per system and oblique musical score, an OMR prototype system has been developed and designed. At last, about 50 musical score image which are downloaded from the websites have been used to demonstrate the efficiency and accuracy of the system.