Hello, I have a quick question based on your past experience.
I have a product where I am in control of the form that I will be later processing with your product. Therefore, I have the opportunity to select among the field marking types that are listed here:
My question is, in your experience, which of these field types generally produces the highest quality results? I've tried a few with very mixed quality and would like to maximize the possibility of getting a high quality result.
Preferable marking type is 'greyBoxes'. It gives the least amount of artifacts during binarization, a process of making bi-tonal image used by recognizer.
An alternative to gray boxes is boxes with 'dotted' borders. It is not presented in API yet but 'simpleText' type can be used instead. In case of 'dotted' borders they will be removed by de-speckling filter before recognition.
Best regards, Dmitry.
answered 22 Aug '12, 12:21