RE: image registration & comparison

The concept of comparing an image with a master image is highly sensitive to
slight errors in registration. It is also highly sensitive to slight errors
that are of no importance. As a result, it is generally not a good way to
detect manufacturing defects. A more practical approach is to categorize the
types of features that appear on images (good and bad). Then have a pass
that recognizes and locates all the features. Then compare the list of
actual features with those that were expected. When you get a match to
within say 1 pixel, it is good; otherwise it is bad.

But perhaps your application is sufficiently different that my experience
isn't relevant to it. Good luck!

-----Original Message-----
From: OpenCV@xxxxxxxxxxxxxxx [mailto:OpenCV@xxxxxxxxxxxxxxx] On Behalf Of
Lorenzo Monti
Sent: Tuesday, January 31, 2006 2:56 AM
To: OpenCV@xxxxxxxxxxxxxxx
Subject: [OpenCV] image registration & comparison

Dear all,
I'm new in image analysis, so I apologize for my
(maybe) stupid questions.
I'm facing this project:

1) acquire an image from a scanner (TWAIN)
2) register it (rotate & shift)
3) match it against a "master" considered to be "good"
  (pixel by pixel difference of the two images,
   then gradient of the result to detect big
differences)
   
The goal is automatic recognition of defects on a printed film.
Before going on, I'd like to hear some experienced user's opinion:
- is OpenCV the right choice for jobs 2 & 3?
- are OpenCV edge detection routines precise enough to grant
  perfect alignment between the scanned image and the master?
- am I on the right way? I don't feel confident expecially about point 3.
  read: how would you solve this problem if you were in my shoes?
- could someone point me out on some literature about images comparison?
Thank you for your answers,
Lorenzo



        

        
                
___________________________________
Yahoo! Mail: gratis 1GB per i messaggi e allegati da 10MB
http://mail.yahoo.it




Change settings: http://www.yahoogroups.com/mygroups, select
   Get Emails (get all posts)
   Daily Digest (one summary email per day)
   Read on the web (read posts on the web only)Or Unsubscribe by mailing
OpenCV-unsubscribe@xxxxxxxxxxxxxxx
 
Yahoo! Groups Links



 






Change settings: http://www.yahoogroups.com/mygroups, select
   Get Emails (get all posts)
   Daily Digest (one summary email per day)
   Read on the web (read posts on the web only)Or Unsubscribe by mailing 
OpenCV-unsubscribe@xxxxxxxxxxxxxxx