A New Approximation Method of the Quadratic Discriminant Function
Proceedings of the Joint IAPR International Workshops on Advances in Pattern Recognition
ICDAR '05 Proceedings of the Eighth International Conference on Document Analysis and Recognition
Discrete Curve Evolution Based Skeleton Pruning for Character Recognition
ICAPR '09 Proceedings of the 2009 Seventh International Conference on Advances in Pattern Recognition
Handwritten character recognition through two-stage foreground sub-sampling
Pattern Recognition
Isolated Handwritten Malayalam Character Recognition Using HLH Intensity Patterns
ICMLC '10 Proceedings of the 2010 Second International Conference on Machine Learning and Computing
Hi-index | 0.00 |
In this work, we have experimented with a simple feature for Malayalam hand written character recognition. Character images are first divided into zones and feature vector is formed by traversing each zone diagonally. For the classification, a Simplified Quadratic Classifier (SQDF) is used. The study was carried out with a database containing 19,800 isolated handwritten characters pertaining to 44 classes. We have obtained a recognition accuracy of 97.6% with SQDF -- Diagonal Feature pair for k = 11, with a feature vector of size 54. It is found to be the best result reported in Malayalam HCR. For comparison, we have used the well accepted gradient feature. The highest recognition rate obtained with gradient feature is 95.24%. As the diagonal based feature is simple, this is a remarkable achievement in Malayalam Handwritten Character recognition.