The most recent sub-project of the most recent project (FOREX Analyzer) of mine (as of February 19, 2003). I had to develop a precise (no errors) OCR. If you ever think about writing an OCR program, this could be the core of the program.
This is exactly what i was searching for
i have some problems though
first,
I'm using vb.net,not vb6
so some of the classes u r using not being used there
the reason why i'm using vb.net cuz i'm developing the application for windows mobile,to be used for pocket pc devices
can u help me out plz?
How can i put an image into it and get it to recognise the text. Every time i try, the image turns to a muddle of black and white pixels, then goes back to normal without any recognsed text coming up in the textbox.
Sorry, forgot to say also that if i open word, write something, take screenshot, and import that, it just decyphers it as naughts. Lastly, how do i work out number of pixels a letter takes up.
Well the OCR is 100 percent accurate. Any alternations, especially if you are using a different font, not to mention the size is even different, the code will not work. Inconvinient? Yes, but darn accurate. This code is just an example, you would have to derive your own 100% accurate code for your own purposes using my technique (or any other you may find suitable). If you want to make an OCR that will work with anything "reasonable", you will increase the chance for mistakes. For example simple letters like O and D sometimes on some fonts could be easily confused even by a naked eye, let alone an OCR that calculates percent similarity and gives best match.
then surely it's not a true ocr. OCR is designed to get text from any font, size, etc, even handwriting sometimes.
KAZAR
The Law Of Programming:
As the Number of Lines of code increases, the number of bugs generated by fixing a bug increases exponentially.
__________________________________ www.startingqbasic.co.uk
No it's not the OCR you usually see that works with most fonts, sizes, etc....this is an example OCR for Arial 9 font that is more accurate than the OCR you are talking about. Again, it works just for ARIAL 9 font, you would need to completely rewise the technique or most likeley use a different method for usual OCR type that works with a lot more than Arial 9 but it's also not 100% accurate, at least on some fonts.
heres a scenario.. i know how to add letters to the array and all that good stuff.. lets say i'm reading an image that has "Lv" in it, there is no 1 pixel space between the letters thus it would not return anything at all.. how would that be delt with?
(as of February 19, 2003). I had to develop a precise (no errors) OCR. If you ever think about writing an OCR program, this could be the core of the program.
Reply With Quote
