Locating segmentation regions of connected handwritten digits
Segmentation of handwritten touching characters is an important research topic in Optical Character Recognition (OCR). Finding the initial candidate segmentation points is the first important step. A method is described to locate the initial candidate segmentation regions based on the boundary analysis of a character string. This method can reduce the redundant searching regions significantly. Therefore, the segmentation reliability can be increased and the segmentation time can be reduced.
Unable to display preview. Download preview PDF.
- 3.Casey, R. G., Nagy, G.: Recursive segmentation and classification of composite patterns. in Proc. Sixth Int'l Conf. Pattern Recognition, (1982) 1023–1026Google Scholar
- 5.Bayer T. A., Kreßel, U. H.-G.: Cut classification for segmentation. in Proc. Int'l Document Analysis and Recognition, (1993) 565–568Google Scholar
- 8.Strathy, N. W., Suen, C. Y., Krzyzak, A.: Segmentation of handwritten digits using contour features. in Proc. 2nd Int'l Conf Document Analysis and Recognition, (1993) 577–580Google Scholar
- 11.Hu, J., Yu, D., Yan, H.: A multiple point boundary smoothing algorithm. Pattern Recognition Letters, accepted.Google Scholar