Pattern Matching (general question)

asked 2015-03-24 21:49:42 -0500

KansaiRobot gravatar image

updated 2015-03-24 21:52:03 -0500

Hello and thanks always for your help.

This time, I would like to ask a rather general question, please forgive the generality. I have been doing segmentation and morphology operations, watershed etc so far. Now I would like to study the topic of Pattern Matching and how to do it in OpenCV.

Now, I have done pattern matching using other tools, and I am just opening my OpenCV books to look for something similar. What should I be looking for in them, I mean what terminology do OpenCV use for this??

(Note: I understand pattern matching as the process of creating a geometrical model (say a circle, or a elongated one) and find instances of these in the image -with varying sizes and orientations.)

Thanks for any comments, pointers or advice. :)

edit retag flag offensive close merge delete



basically you should have a look for terms like template matching, object detection/matching using different kind of feature keypoints algorithms. These are the simple ones, if you want something more advanced then you should have a look in other more complex machine learning algorithms and techiques.

theodore gravatar imagetheodore ( 2015-03-24 22:05:47 -0500 )edit

Thank you for your comment. I have checked so far only matchTemplate and although it fits what I am looking for to some extent, I guess this is a brute force approach and sensible to scaling and rotation?? I am looking for some method that finds several ocurrences of a synthetic model which can be in different scales or rotated.... I guess I will keep looking :)

KansaiRobot gravatar imageKansaiRobot ( 2015-03-26 04:02:04 -0500 )edit

@KansaiRobot if you search about template matching I thing you will find examples where scaling and rotation is introduced. Of course in these cases the computing time increases. For the latter you could have a view here, about how you can decrease the computing time.

theodore gravatar imagetheodore ( 2015-03-26 09:06:42 -0500 )edit