An OMR based automatic music player

Abstract

Automated learning systems used to extract useful information from musical scripts, play a major role in optical music recognition. Optical music recognition or OMR has been widely used to extract the musical notations and knowledge from old scripts and thus enclose lot of importance in retrieving historical data. The field of pattern recognition and knowledge representation has to be symmetrically used for music notation recognition making it a challenging problem. Most of the methods used for OMR recognition and extraction like HMM's, Neural etc mentioned in literature have errors which require human operators to be rectified. The paper proposes a novel automatic music recognition system which can be effectively used to recognize sheet or printed scores into a playable platform under the different conditions of blur, illumination variations, noise etc. The paper uses an enhancement supported threshold GA based pre-processing methodology for bimodal images further compares with the Sobel filtered and standard GA outputs for the same. The methodology makes use of an enhanced image obtained by histogram equalization followed by image segmentation using a specific threshold. The threshold can be obtained using genetic algorithms. GA based segmentation technique is codified as an optimization problem used efficiently to search maxima and minima from the histogram of the image to obtain the threshold for segmentation. The system described is capable of extracting normal images as well as images with noise, images from old scripts.