乐谱识别中的结构拆分及图象恢复  

Structural Decomposition and Image Reconstruction in Music Recognition

作　　者：张登胜[1] 徐秉铮[2] 马颂德[3] 

机构地区：[1]安庆师范学院数学系,安庆246011 [2]华南理工大学无线电所,广州510641 [3]中科院自动化所模式识别国家重点实验室,北京100080

出　　处：《数据采集与处理》1997年第3期238-242,共5页Journal of Data Acquisition and Processing

基　　金：安徽省自然科学基金

摘　　要：在电子出版及许多音乐研究中，均需将原文乐谱转化为被计算机可读的数据。本文提出一个用字符识别方法的乐谱图象识别系统。它基于拆分结构技术，将原文转换为局部结构图切分乐符，有效地减少了数据量，且不受图象弯曲和倾斜的影响。文中给出了识别结果的文本输出及其图象恢复。In electronic publishing and many music software, there is a need to trans form music score to computer readable data. In this paperl an automatic recognition system for printed music is given with OCR method. Based on structural decomposition technique, it segments music symbols by transforming the original music score into local structural graph. This results in significant data reduction and it is insensitive to curvature and rotations. A recognition output is given in the form of ASCll file and the original image is reconstructed.

关 键 词：结构拆分 图象恢复 乐谱识别 图象处理 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]