Modified Quadratic Discriminant Functions and the Application to Chinese Character Recognition
IEEE Transactions on Pattern Analysis and Machine Intelligence
Decision Combination in Multiple Classifier Systems
IEEE Transactions on Pattern Analysis and Machine Intelligence
IEEE Transactions on Pattern Analysis and Machine Intelligence
DAS '02 Proceedings of the 5th International Workshop on Document Analysis Systems V
Hybrid Pen-Input Character Recognition System Based on Integration of Online-Offline Recognition
ICDAR '99 Proceedings of the Fifth International Conference on Document Analysis and Recognition
Robust and Highly Customizable Recognition of On-Line Handwritten Japanese Characters
ICPR '96 Proceedings of the International Conference on Pattern Recognition (ICPR '96) Volume III-Volume 7276 - Volume 7276
IWFHR '02 Proceedings of the Eighth International Workshop on Frontiers in Handwriting Recognition (IWFHR'02)
Two On-Line Japanese Character Databases in Unipen Format
ICDAR '01 Proceedings of the Sixth International Conference on Document Analysis and Recognition
Collection and Analysis of On-line Handwritten Japanese Character Patterns
ICDAR '01 Proceedings of the Sixth International Conference on Document Analysis and Recognition
Direction-Change Features of Imaginary Strokes for On-Line Handwriting Character Recognition
ICPR '98 Proceedings of the 14th International Conference on Pattern Recognition-Volume 2 - Volume 2
ICDAR '05 Proceedings of the Eighth International Conference on Document Analysis and Recognition
IAM-OnDB - an On-Line English Sentence Database Acquired from Handwritten Text on a Whiteboard
ICDAR '05 Proceedings of the Eighth International Conference on Document Analysis and Recognition
Hi-index | 0.00 |
This paper presents a technique for normalizing likelihood of multiple classifiers, allowing their fair combination. Our technique generates for each recognizer one general or several stroke-number specific characteristic functions. A simple warping process maps output scores into an ideal characteristic. A novelty of our approach is in using a characteristic based on the accumulated recognition rate, which makes our method very robust and stable to random errors in training data and requires no smoothing prior to normalization. In this paper we test our method on a large database named Kuchibue_d, a publicly available benchmark for on-line Japanese handwritten character recognition and very often used for benchmarking new methods.