I want to segment offline handwritten Arabic words into characters. Here is an attached file for examples of offline Arabic handwritten words (persons' names). Which algorithm may I follow to achieve this? Is it suitable to use the vertical and/or horizontal projection? How can I do that using MatLab? Or may I use the histogram approach? Also, How can I implement this in Matlab? Please, I need an easy and clear approach so I can follow it (I am a novice).
Dear Nisreen,
What i introduce here may not be an exact answer but some solution thoughts to the problem.
- You can identify the hand written words by comparing them with their tempelates
- So you need to build data base for word templates
- To identify a new hand written word you compare it with word templates in the data base and get out the highest matched one. The best matched word is the targeted one.
- These words are written in ASCII format
- Then it is easy to get out its characters.
The matching can be accomplished by calculating the correlation between the intended word and the templates.
- One can use neural networks for the identification process to save time and increase the accuracy.
- There are matlab tool for signal processing that can be applied to solve this problem
Best wishes
@ Abdelhalim thank you for your replying. This is a good idea, I think it works with limited words (only that are in template data base). I will follow your above steps. If it possible, I need a useful link to start implementing those steps. Thank you much.
Dear Nisreen,
Please look at the paper in the link:https://pdfs.semanticscholar.org/9be0/69d4f9241781825ab701b914e919be1ed0fb.pdf
Best wishes
- Badges
- Science topic
- Similar topics
- Filing