Machine Vision

cancel
Showing results for 
Search instead for 
Did you mean: 

Using OCR on Rotated images

I am trying to build a VI that will perform OCR on a image. I would like to make the VI robust enough to handle slightly rotated images. The problem I am encountering is that the region of interest can not be drawn at an angle. Is there a way to establish a coordinate system and draw the roi referencing this new coordinate system. Any suggestions would be great
0 Kudos
Message 1 of 6
(4,848 Views)
You should be able to use the ROI Descriptor for the position correction of the read function. I created an example which do that using LabVIEW 7.1 and Vision 7.1. You can find an screenshot attached. I search for a pattern to get the new angle in the run time image before using the read function.
0 Kudos
Message 2 of 6
(4,848 Views)
What do the two Set VIs and the Update VI do? Thanks for the help.
0 Kudos
Message 3 of 6
(4,848 Views)
1. I use the set VI to learn a pattern match and set a coordinate system.
2. In the next SubVI I select an ROI for the Read OCR function. The reference of the ROI coordinate is the position of the new coordinate system.
3. The first update SubVI searches the pattern and set the coordinate system again.
4. The second function in the while loop reads the text on the right position (because I work with the new coordinate system),again.
Step 1. and 2. are the teach in for this appliation.
Step 3. and 4. are the real machine vision inspection.

I can attach my llb, if you are using the current versions of LabVIEW and Vision Development Module (7.1). You will get the current NI OCR Software automatically with Vision since version 7.1.
0 Kudos
Message 4 of 6
(4,848 Views)
I would very much like to look at your library file but I only have LV 7.0 and Vision 7.0. Thanks for the help though you have given me some ideas.
0 Kudos
Message 5 of 6
(4,848 Views)

Dear Steffen,

 

I am facing the same problem. The vi images you posted couldn't help much. Could you post your vi for better understanding.

 

Thank you.

 

Hart 

0 Kudos
Message 6 of 6
(4,141 Views)