[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: SIRAGI, I need help !



Can you send me on mirza dot pasovic at uni-mb dot si a sample image file.

THX

On Tue, 2005-05-17 at 15:44 +0000, Tarik FDIL wrote:
> Hi,
> 
> I would like to talk you about a little technical problem which blocked me for 
> several days... 
> 
> You know that OCR processing proceed like this :
> 
> 1- read image file
> 2- detect lines
> 3- detect characters in a line
> 4- compare characters detected with a database of stored characters and decide 
> which characters are related.
> 5- write text resulting from OCR and make some extra processing (spell 
> checking, etc.)
> 
> My technical concern is between 3 and 4 stage. After detecting character's 
> pixels and before sending them to the comparison system, you should make some 
> adjustment to charcter's image :
> - we should clean it (remove isolated pixels, etc.)
> - we should squlettisize the character, ie. transforming it to become thin : a 
> shape made with only one-pixel drawings. This enhance the effectiveness of 
> comparison.
> - we should resize (resample) the charcter to make it fitting in a one 
> standard fixed matrix of pixels.
> 
> I have problem to write a good program for squlettisation of characters 
> (thinning). I tried many methods but I miss documents describing in details 
> thinning algorithms. You can find joind to this mail some programs that I 
> have written. But I'm not satisfied by the output of these programs. Either 
> they didn't respect the overall shape of character, or they eliminate too 
> many pixels, etc.
> 
> So, if some one have documents, librairies or programs which can help us in 
> "shape thinning or squelettisation", it will be very welcome.
> 
> Best regards 
> Tarik
> 
> _______________________________________________
> Developer mailing list
> Developer at arabeyes dot org
> http://lists.arabeyes.org/mailman/listinfo/developer