Online Handwritten Character Recognition

1
On-line Handwritten Character Recognition

• Proposer Dr. S. V. Rao (Co-Investigator)
• Institute Indian Institute of Technology
  Guwahati
• Languages Assamese/Bangla
• Component On-line handwritten character
  recognition system for
  Assamese/Bangla

2
Strength of our Organization

• Assamese OCR
• Developed OCR systems for Assamese (with
  collaboration with ISI Calcutta under TDIL-I
  project)
• Developed Layout analysis.
• Developed postprocessor.

3
On-line handwritten OCR

• On-line handwritten OCR for Devanagari on Pocket
  PC (work in progress)
• System recognizing basic characters and matras.
  Around 90 accuracy.
• Work is going on for half characters and
  Juktakshars.
• Implemented on Toshiba e-800 series.
• Method Hybrid approach.

4
(No Transcript)
5
Font Independent OCR

• Proposer Dr. S. V. Rao (Co-Investigator)
• Institute Indian Institute of Technology
  Guwahati
• Languages Assamese/Bangla
• Component Font independent OCR for
  Assamese/Bangla

6
Limitations of Assamese OCR(Developed under TDIL
Phase I)

• Font dependent
• Training needed for new fonts
• Size of database
• Increases as number of fonts increase.
• Speed of recognition depends on the size of
  database.