International Conference on Computer Engineering and Technology, 3rd (ICCET 2011)
20 Lexicon Reduction Using Delayed Strokes for the Recognition of Online Farsi Subwords
Download citation file:
- Ris (Zotero)
- Reference Manager
In this paper, we investigated the efficiency of delayed strokes like dots on lexicon reduction and also increasing the recognition accuracy in online Farsi handwriting. Many Farsi letters in a word have peripheral and delayed strokes. Two filters relying on the delayed strokes were proposed to reduce the lexicon. These filters were based on location of the delayed strokes and their sequence. The proposed lexicon reduction was implemented on TMU dataset containing 1000 handwritten Farsi subwords. Experimental results yielded the speedup factor of 3 as well as 15% improvement in the recognition rate.