Machine Vision

cancel
Showing results for 
Search instead for 
Did you mean: 

Vision software not always reading numbers and converting proper image to text

Hello,

I have am using vision code to convert image to text. Basically takes a picture of an image on a card and within the card there are two row of numbers.  I am only interested in a particular line of numbers (different font) and has a total of 20 numbers.  The other has less numbers and different font.  The camera I am using is a basler acA2500-14gm.  There are a total of 6 cards spaced about 3-5 inches apart.  There are 3 cameras. One camera has 2 cards in its FOV.  There is also lighting that was recommended by the distributor for basler.

 

The logic of code basically is as follows:

Setup: 

I used the vision tool to learn the set of numbers 0-9.  I also have a learned template for each card, specifically the row of number. Each card has slight difference in light intensity since the light hits the closer card more than the farther one.

code logic:

1- Take a picture and rotate image 90deg. I rotate the image because it is easier for user to read in case the converted image to text is incorrect.

2- load image template

3- pattern match

4 - use OCR vi (modified it to suite my needs) to convert image to text.

Problem:

Sometimes it doesn't place the small square on the numbers.  Sometime not all the numbers have small squares on the numbers.

Sometimes the pattern match isn't found or the pattern is nowhere close to the pattern.

0 Kudos
Message 1 of 5
(2,793 Views)

If you can attach some images, that would be very helpful.  It is hard to give advice without being able to see the problems you are having.

 

Bruce

Bruce Ammons
Ammons Engineering
0 Kudos
Message 2 of 5
(2,760 Views)

Hello Bruce,

Thanks for the reply.  Unfortunately, my boss has decided to move in a different direction. If I work on the project again, I will take pictures of what I am talking about.

 

thanks again,

Jose

0 Kudos
Message 3 of 5
(2,555 Views)

Hey, is it what you said similar with the scan mechadism in which it can transfer image of text to text?

0 Kudos
Message 4 of 5
(2,533 Views)

Hey Jessica,

 

I think what you are asking is if OCR is what you use to translate a picture of text into a string of text that can be used programmatically.

 

If that is the question then yes.  Check out this example to see what you can do with OCR!  Also, I noticed you just joined.  If you have a question or a topic you'd like to discuss, I'd recommend starting a new thread in the forums so that we keep each thread to one topic.

 

-Bear

Regards,

Ben Johnson
ʕง•ᴥ•ʔง
0 Kudos
Message 5 of 5
(2,522 Views)