Recognition of handwritten digits using template and model matching
作者： Paul Gader , Brian Forester , Margaret Ganzberger , Andrew Gillies , Brian Mitchell
DOI: 10.1016/0031-3203(91)90055-A
关键词: Artificial intelligence 、 Pipeline (computing) 、 Speech recognition 、 Computer science 、 Model matching 、 Numeral system 、 Pattern recognition 、 Test data 、 Pattern recognition (psychology) 、 Template matching
摘要: Abstract A pipeline strategy for handwritten numeral recognition that combines a two-stage template-based technique and model-based is described. The template matcher multiple information sources. second stage of the was trained on rejects from first stage. classifies 70–80% digits with reliability rates over 99%. It also generates class membership hypotheses remaining which constrain system. Recognition 94.03–96.39% error 0.54%–1.05% are obtained test data consisting 13,000 well-segmented ZIP codes in USPS mail.
elsevier.com
acm.org 
sci-hub.se 参考文章(3)
R. G. Casey, Moment Normalization of Handprinted Characters IBM Journal of Research and Development. ,vol. 14, pp. 548- 557 ,(1970) , 10.1147/RD.145.0548
Louisa Lam, Ching Y. Suen, Structural classification and relaxation matching of totally unconstrained handwritten zip-code numbers Pattern Recognition. ,vol. 21, pp. 19- 32 ,(1988) , 10.1016/0031-3203(88)90068-4
J.D. Tubbs, A note on binary template matching Pattern Recognition. ,vol. 22, pp. 359- 365 ,(1989) , 10.1016/0031-3203(89)90045-9