Home List your patent My account Help Support us

OCR METHODS FOR IDENTIFYING UNCLEAR CHARACTERS

[Category : - SOFTWARES]
[Viewed 1440 times]

OCR SOFTWARE WITH METHODS FOR IDENTIFYING UNCLEAR CHARACTERS

Firstly there is a standard format for most printing. And there are limited in number. That means that if we know the gap between words, we can predict the average size or space that each letter will occupy. That is because if we know the inter word gap size, and only one type of print system has that inter word gap size, then we can find out using its specification ,what size of letters and the inter character space it uses is.

This is found by detecting the size of the gaps between the words, finding their average and comparing in a database of values obtained from these writing formats. So each printing systems information will be included in the database of the software. Once we know the letter space (using above method) we go through the unclear word chopping it up into those segments that should contain complete letters.

.

We know that as we focus the image being read , the characters get blurry and unclear.

This does not have to be a problem.

An unclear "W" is blurred differently from an unclear "i", though the difference in appearance is less than in a clear image. So once we know the letter space (using above method) we go through the unclear word chopping it up into those segments that should contain complete letters.

Then the unclear characters are compared to a database of unclear characters. Those in the database will be know off, off which letter they are blurred and so after matching , the correct character can be recognized.

We can make it even more accurate.

When using a camera to read a document.

As the characters gets into focus the pixels of the image arrange themselves differently from those in the unclear image to those of the clear character. We then track how they migrate. An unclear "W's" pixels will move in a different direction to a clear “W”s” differently from how unclear “I’s” pixels will. so we can record this movement and compare that to a database of movements that are known.

Patent publications:

No publication

Asking price:

Make an offer

Rate this patent

Great invention
Viewed: 1440 times

Seller:

moyo (Zimbabwe)

Ask a question
or start a negotiation

Contact the seller

Share on social media