Screenshots     Download     Pricing     Documentation     Tutorials     Resources     Contact     Forum     Search  

This Thread

 New Post   Forum Index 

 
Shape Matching Issue
Jeff McGrew  [4 posts]
4 years
I am trying to use the shape matching module to recognize shapes, however want it to ignore extra pixels beyond the shape I am wanting to recognize.   I created a simple circle, and added an extra line.  I have to drop the confidence level to 66 percent to recognize the shape.  Is there a way to accomplish this?  The shapes I am trying to recognize have extra pixels, and I would like to be able to recognize the basic shape without having to include the extra pixels.  Also must the shape being recognized be one connected body without any seperations?

 
Anonymous 4 years
Jeff,

You may try to pre-process the shape prior to identification. For example, you can use the Smooth_Hull module to remove extra pixels prior to recognition.

http://www.roborealm.com/help/Smooth_Hull.php

Or you could alternatively try the erode module which does something similar

http://www.roborealm.com/help/Erode.php

Or if you shapes are really circles then using the canny edge detection followed by the circle detection module you would be able to track circles.

http://www.roborealm.com/help/Canny.php
http://www.roborealm.com/help/Circles.php

Yes, the shapes need to be separate objects before being recognized. You can try using the erode module to help with this. Also if the connections are too large you can try the Watershed or the Blob_Split module to disconnect shapes.

http://www.roborealm.com/help/Blob_Split.php

If these do not help post the images that are causing difficulty and we can take a look.

STeven.

This forum thread has been closed due to inactivity (more than 4 months) or number of replies (more than 50 messages). Please start a New Post and enter a new forum thread with the appropriate title.

 New Post   Forum Index 


© 2005 - 2014 RoboRealm. All Rights Reserved. | Contact | Glossary | Privacy | Disclaimer | Link to Us | Resources | Site Map