Visual music score detection with unsupervised feature learning method based on K-means
- PDF / 2,751,340 Bytes
- 11 Pages / 595.276 x 790.866 pts Page_size
- 31 Downloads / 187 Views
ORIGINAL ARTICLE
Visual music score detection with unsupervised feature learning method based on K-means Yang Fang • Teng Gui-fa
Received: 7 November 2013 / Accepted: 22 April 2014 Ó Springer-Verlag Berlin Heidelberg 2014
Abstract Automatic music score detection plays important role in the optical music recognition (OMR). In a visual image, the characteristic of the music scores is frequently degraded by illumination, distortion and other background elements. In this paper, to reduce the influences to OMR caused by those degradations especially the interference of Chinese character, an unsupervised feature learning detection method is proposed for improving the correctness of music score detection. Firstly, a detection framework was constructed. Then sub-image block features were extracted by simple unsupervised feature learning (UFL) method based on K-means and classified by SVM. Finally, music score detection processing was completed by connecting component searching algorithm based on the sub-image block label. Taking Chinese text as the main interferences, the detection rate was compared between UFL method and texture feature method based on 2D Gabor filter in the same framework. The experiment results show that unsupervised feature learning method gets less error detection rate than Gabor texture feature method with limited training set. Keywords Visual image Music score Unsupervised feature learning Texture Gabor
1 Introduction Optical music recognition (OMR), which transforms digital music score image to computer readable format symbols, is very important for automatic music image processing. As the basic process of OMR, music score detection is to determine the locations and boundaries of music scores in the whole image. Currently, with the development of electronic portable vision equipments, such as mobile telephone and digital camera, more and more music score images are conveniently taken as visual images. The visual music score images have become the main object in OMR research area [1]. Compared with the images taken by traditional flat-bed scanners, the visual music score images, especially those taken from a book, have some new characteristics which bring more difficulties of recognition (just as shown in Fig. 1): 1.
2.
3. 4.
Y. Fang T. Gui-fa (&) College of Mechanical and Electrical Engineering, Agricultural University of Hebei, Lingyusi Street, No. 289, Baoding, China e-mail: [email protected] Y. Fang College of Mathematics and Computer Science, Hebei University, Wusi East Road, No. 180, Baoding, China
The staff lines in music scores often have various kinds of severe deformations, such as curvature, staggered, lens deformation, etc. The color and brightness of images are not uniform caused from the shooting conditions variation such as different light sources. There are many other objects or complex backgrounds around the music score. Music note definitions are not consistent since the variable shooting distance and camera resolution ratio.
As the staff-lines are the obvious chara
Data Loading...
