Training Set Expansion in Handwritten Character Recognition
In this paper, a process of expansion of the training set by synthetic generation of handwritten uppercase letters via deformations of natural images is tested in combination with an approximate k-Nearest Neighbor (k-NN) classifier. It has been previously shown   that approximate nearest neighbors search in large databases can be successfully used in an OCR task, and that significant performance improvements can be consistently obtained by simply increasing the size of the training set. In this work, extensive experiments adding distorted characters to the training set are performed, and the results are compared to directly adding new natural samples to the set of prototypes.
KeywordsSynthetic Image Local Database Neighbor Rule Handwritten Character Neighbor Search Algorithm
- 3.P. A. Devijver and J. Kittler. On the edited nearest neighbour rule. In Proceedings of the 5th International Conference on Pattern Recognition, pages 72–80. IEEE Computer Society Press, Los Alamitos, CA, 1980.Google Scholar
- 5.T. M. Ha and H. Bunke. Off-line, handwritten numeral recognition by perturbation method. IEEE Trans. on PAMI, 19(5):535–539, May 1997.Google Scholar
- 7.A. Jain. Object matching using deformable templates. IEEE Transactions on Pattern Analysis and Machine Intelligence, 18:268–273, 1996.Google Scholar
- 10.J. C. Perez-Cortes, J. Arlandis Arlandis, and R. Llobet. Fast and accurate handwritten character recognition using approximate nearest neighbours search on large databases. In Workshop on Statistical Pattern Recognition SPR-2000, Alicante (Spain), 2000.Google Scholar
- 11.S. J. Smith. Handwritten character classification using nearest neighbor in large databases. IEEE Trans. on PAMI, 16(9):915–919, September 1994.Google Scholar