Research Of The Method Of Chinese Pipa Gong-Che Notation Music Score Symbols Recognition

Gong-Che Notation (GCN) is a major notation method of Chinese traditional music, which is inherited as the vector by the music of the most common Chinese traditional musical instruments, such as Pipa GCN. It isn't conductive to the spread and inheritance of traditional Pipa music, because it's paper's work and the notation isn't universal. With the development of the technology of Optical Character Recognition (OCR) and Optical Music Recognition (OMR), Musical Notation, Numbered Musical Notation (NMN) and Chinese documents have realized the digital conversion, which lays technical basis for the digital music conversion of Pipa GCN. The score symbols recognition of paper's Pipa GCN is studied, in this paper, and which has the great significance for the spread and inheritance of Chinese traditional music in the digital context.Based on the comprehensive analysis of music recognition at home and abroad and digital realization technology, the key technologies and methods in the score symbols recognition processes of Pipa GCN are discussed. The realization methods for score symbols recognition of Pipa GCN are established depending on the notation features and the music image characteristics. In the score symbols extraction method of Pipa GCN, a new kind of music segmentation algorithm basing on the Pixel-Tracing method is presented in this paper. By means of counter- clockwise rotation extension scanning, the smallest external rectangle can be labeled surrounding each pixel connected region which is the component of the music sore symbols. This algorithm not only makes the extracted score symbols or their parts cleaner and more precise, but also breaks through the traditional model of image segmentation based on stroke, reduces the segmentation post-processing. In the score symbols feature extraction method of Pipa GCN, uniform grid method is used to extract score symbols features, and the algorithm is simple and easy to realize. In the score symbols classification and recognition method of Pipa GCN, Neural Network including BP and RBF models is used to design classifier for score symbols recognition of Pipa GCN. Furthermore, this article compares and analyzes computing performance and recognition accuracy of the score symbols classifiers designed by the two models above. Finally, the classifiers are improved according to the feature of score symbols.The experimental system for score symbols recognition of Pipa GCN is built and developed, in this paper, and which includes three Modules: pre-processing,feature extraction and classification and recognition. The results of recognition from the experimental system show that the system performs well and has quick recognition speed, besides, it has better recognition effect.