Old Handwritten Music Symbol Recognition Using Directional Multi-Resolution Spatial Features

Automatic recognition of musical symbols received huge attention in the last two decades. Most of the work is carried out for the recognition of printed symbols whereas little attention is given to handwritten symbols. In handwritten musical symbols, when we deal with historical and old handwritten musical symbols, the problem becomes more challenging. In this paper, we have dealt with recognition of old handwritten musical symbols. In our method, we have used directional multi-resolution statistical descriptors by combining Radon Transform, Discrete Wavelet Transform, and Statistical Filters. Simple k-NN classifier is used with fivefold cross validation. We have achieved encouraging results on our dataset.