Dynamically Expanding Context as Committee Adaptation Method in On-Line Recognition of Handwritten Latin Characters.
Jorma LaaksonenMatti AkselaErkki OjaJari KangasPublished in: ICDAR (1999)
Keyphrases
- recognition algorithm
- dynamic programming
- preprocessing
- high recognition accuracy
- similarity measure
- significant improvement
- computational cost
- high accuracy
- neural network
- pattern recognition
- pairwise
- character segmentation
- recognition rate
- detection method
- segmentation method
- edge detection
- adaptation process
- handwritten characters
- tree grammars
- recognition accuracy
- high precision
- cost function
- video sequences
- feature extraction