Writer identification from musical score documents is a challenging task dueto its inherent problem of overlapping of musical symbols with staff lines.Most of the existing works in the literature of writer identification inmusical score documents were performed after a preprocessing stage of stafflines removal. In this paper we propose a novel writer identification frameworkin musical documents without removing staff lines from documents. In ourapproach, Hidden Markov Model has been used to model the writing style of thewriters without removing staff lines. The sliding window features are extractedfrom musical score lines and they are used to build writer specific HMM models.Given a query musical sheet, writer specific confidence for each musical lineis returned by each writer specific model using a loglikelihood score. Next, aloglikelihood score in page level is computed by weighted combination of thesescores from the corresponding line images of the page. A novel Factor Analysisbased feature selection technique is applied in sliding window features toreduce the noise appearing from staff lines which proves efficiency in writeridentification performance.In our framework we have also proposed a novel scoreline detection approach in musical sheet using HMM. The experiment has beenperformed in CVC-MUSCIMA dataset and the results obtained that the proposedapproach is efficient for score line detection and writer identificationwithout removing staff lines. To get the idea of computation time of ourmethod, detail analysis of execution time is also provided.
展开▼
