Re: When I OCR a complete document a number of times I receive different results

I am developed a crucial application which in summary scans invoices of numerous formats and selects specific key data  based upon key words. The process is relatively simple by scanning the array of recognised words for user definable word patterns and either selecting other words offset from them or between a range of word patterns. This part of the project seems to be fine whereby the user builds up an invoice profile so when an invoice of the same format is scanned a second time the key data can be recognised and selected from the profile set up. This would be fantasic however if I scan the same document a number of times it invariably returns the recognised word array in a different order thus rendering the profile as invalid. My code for performing ocr on the document is virtually the demo code supplied with version 14.5 I am not setting manual zones or anything complicated. I have spent the morning scanning the forum for similar issues with no luck in finding a solution. The success of this proto-type project will determine whether lead-tools ocr will be built in to my applications or another tool so could you please help.

Regards

 

Alban

 

The list of recognized words in NOT given in any specific order. However, it includes the X and Y coordinates of each word. This means if you want to order then based on location, you should be able to do that.


Thanks,
Ali Abo Al-Rob
LEADTOOLS Technical Support