How can I match a masked image using matchTemplate() in OpenCV with C++?
This is my masked template image:
And this is my source image:
Asked
Active
Viewed 2,820 times
0
-
2May be this helps you: http://stackoverflow.com/questions/4761940/opencv-template-matching-and-transparency – Barshan Das Apr 16 '13 at 07:33
2 Answers
2
matchTemplate's fifth argument is a mask array which you can use for this purpose. In your case, you'll want a binary mask. That is, a Mat with:
- depth
CV_8U, and - dimensions equal to your template image, and
- the pixels you want to ignore from your template set to 0, and
- the pixels you want to be used set to a non-zero value (typically 1)
Mark Amery
- 127,031
- 74
- 384
- 431
0
Look at opencv example. You gotta call the matchTemplate function:
matchTemplate( img, templ, result, match_method );
Safir
- 882
- 7
- 9
-
@Sadir: I looked at that example and I used matchTemplate() before, but I would like to know the best way to match a template image that is masked. – Jaime Ivan Cervantes Apr 16 '13 at 06:31
-
I don't understand what you mean by > "I would like to know the best way to match a template image that is masked". – Safir Apr 16 '13 at 06:33
-
If you look at the template image in my question you will see that it has black pixels. In a masked image, the black pixels will be "transparent", and only the pixels with values > 0 will be taken into consideration when matching. – Jaime Ivan Cervantes Apr 16 '13 at 06:39
-
`matchTemplate` does something like convolution, it slides the windows over the image and applies the convolution. So, I don't think it's possible to specify the template image as "mask" the way you want it. – Safir Apr 16 '13 at 06:57