How to make an appropriate size&spacing in OCR inspection?

[William1225]

Hi,

Since I'm a beginner to the OCR toolkit, hope you tolerate my premature question...

I'd like to take use of the OCR function so that I use the Vision Assistant 2013 to facilitate my task.

It easily failed by defalut setting while I process characters like "xo", "xc"...

If I try to modify the Bounding Rect Width to get the right enclosure of the character, but it failed by the latter character "m"

I think this kind of failure is possible in other cases, so it must have some solutions to deal with this situation.

Could someone give me some suggestion or direction? Very appreciated!!!

[BruceAmmons]

I am guessing the issue is kerning, where the space for one character overlaps the space for the next character.  If you can draw a vertical line and hit both characters, you have kerning.  OCR has always had trouble separating these characters.

The "Separate Char Algorithm" looks like a promising option.  I haven't used this version, so I am not sure what it does.

Bruce

[William1225]

Hi, BruceAmmons,

Thanks to your reply:)

I agree that it is kerning in this case so that I got stuck.

The items in the "Separate Char Algorithm" have been tried by my effort but still not worked.

I'm very confused if I could use OCR toolkit to deal with this kind of failure,

since character separation seemed to be a key point to OCR.

[Lucas.Chen]

I think so, I also very complained about NI did not pay much effert in OCR function. Seperation is a critical step in OCR, so much of the image preprocessing work need to be done by the user.