Abstract
In this paper a writer-independent on-line handwriting recognition system is described comparing the effectiveness of several confidence measures. Our recognition system for single German words is based on Hidden Markov Models (HMMs) using a dictionary. We compare the ratio of rejected words to misrecognized words using four different confidence measures: One depends on the frame-normalized likelihood, the second on a garbage model, the third on a two-best list and the fourth on an unconstrained character recognition. The rating of recognition results is necessary for an unsupervised retraining or adaptation of recognition systems as well as for a user friendly human-computer interaction avoiding excessive call backs. © Springer-Verlag Berlin Heidelberg 2002.
Cite
CITATION STYLE
Brakensiek, A., Kosmala, A., & Rigoll, G. (2002). Evaluation of confidence measures for on-line handwriting recognition. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 2449 LNCS, pp. 507–514). Springer Verlag. https://doi.org/10.1007/3-540-45783-6_61
Register to see more suggestions
Mendeley helps you to discover research relevant for your work.
