Adaptive Thresholding IMAQ Techniques

[JackHamilton]

This paper and sample VI from the NI website is an absolute 10+, a MUST read for all Vision users.

This clever algo - adjusts the background contrast of an image to make it uniform. For pattern rec., OCR or ROI measurements - its a very powerful tool. This will pull those troubled images/lighting problems in rein. Because as we all know - you can't get perfect lighting all the time.

I used this on an OCR problem and like magic! it fixed the problem.

http://zone.ni.com/devzone/conceptd.nsf/we...6256D8F0050B5E3

Cudos to NI! Whoever authored the paper I'll buy a drink for at my next NI Week!

Jack Hamilton

Download File:post-37-1114534365.zip

[yazdihr]

Jack;

I had a quick look at this program. It has a fundamental flaw! (I am surprised that you did not notice that!)

The actual filter size is NOT connected to the input of subvi! Hence, changing the filter size does not change anything. (The default value of 20 is considered!)

I tried various figures and found out that this algorithm is basically kind of enhanced edge detection followed by dynamic histogram stretch and the auro thresholding.

It maybe OK for some application, but it maybe slow for large size filters.

Regards

Hamid

PS. Are you going to be in NI week?

This paper and sample VI from the NI website is an absolute 10+, a MUST read for all Vision users.

This clever algo - adjusts the background contrast of an image to make it uniform. For pattern rec., OCR or ROI measurements - its a very powerful tool. This will pull those troubled images/lighting problems in rein. Because as we all know - you can't get perfect lighting all the time.

I used this on an OCR problem and like magic! it fixed the problem.

http://zone.ni.com/devzone/conceptd.nsf/we...6256D8F0050B5E3

Cudos to NI! Whoever authored the paper I'll buy a drink for at my next NI Week!

Jack Hamilton

4659[/snapback]

[pallen]

Thanks for the pointer to this one Jack!