【作者】 姜一女；

【导师】 叶修梓；

【作者基本信息】 浙江大学 ， 计算机应用技术， 2006， 硕士

【摘要】 简谱是大家非常熟悉和常用的乐谱之一,但是在目前光学乐谱识别领域中对它的研究几乎空白,研究的焦点都集中在五线谱识别上。本文在深入分析简谱特征的基础上,提出了一整套简谱识别系统的实现方法。输入的光学简谱经预处理后,首先通过每行的小节线特征提取出简谱部分,然后通过投影法和种子填充算法定位出简谱符号基元的位置,并由此采用不同的识别算法识别出每个简谱符号基元的类型,最后通过组装把各个简谱符号组装成音乐特征符,形成数字化乐谱。实验表明,这套方法对印刷乐谱的识别达到了令人满意的效果,是一项有意义的研究。更多还原

【Abstract】 Numbered Musical Notation(NMN) is one of the most familiar and common music scores, but it is seldom that anyone studies about its recognition in Optical Music Recognition (OMR) field at present, and everyone focus on the research of Staff Score Recognition. In the base of deeply study of the features of NMN, this paper proposes an approach of fulfilling the recognition. After pre-dealing the inputted optical MNM, extracts the NMN part according to the bar line feature at first. Then locate primitive objects through projections and flood-fill algorithms. And use diverse algorithms to recognize the type of every primitive object. Lastly assemble them to musical features to form digital musical score. The result of the experiment indicate the satisfying effect that achieved by using this method to recognize the NMN.更多还原