In ranorex 2.3 there was a way to do an image recognition to a percentage. I'm having trouble with the image comparison in 3.0 in that two almost identical images are always returning false for comparison or contains. Is there a way to do a similarity comparison based on a percentage as before through the GUI or do I need user code? If so what is an example of user code.
Thanks
image recognition in 3.0
- Support Team
- Site Admin
- Posts: 12145
- Joined: Fri Jul 07, 2006 4:30 pm
- Location: Houston, Texas, USA
- Contact:
Re: image recognition in 3.0
Hi,
We changed a lot in Ranorex 3.0, but this is definitely a part where we changed nothing. The similarity is on the same place as before. If you click the image based action inside your recording, on the right side (if you haven't moved the pad) the properties pad is shown. There is a section called Image Based Location and there you will find the similarity.
Regards,
Peter
Ranorex Team
We changed a lot in Ranorex 3.0, but this is definitely a part where we changed nothing. The similarity is on the same place as before. If you click the image based action inside your recording, on the right side (if you haven't moved the pad) the properties pad is shown. There is a section called Image Based Location and there you will find the similarity.
Regards,
Peter
Ranorex Team
-
- Posts: 169
- Joined: Thu Jan 13, 2011 2:06 pm
Re: image recognition in 3.0
Thanks that helped, though I noticed the image recognition is fairly sensitive. I changed computers to one with a lower resolution and I had to move the recognition percentage from 100% to 95-97% to get some of the images to be recognized. I worry about having to drop the percentage because I don't want "similar" images to pass the test, though I'm not sure of how else this could be solved.
- Support Team
- Site Admin
- Posts: 12145
- Joined: Fri Jul 07, 2006 4:30 pm
- Location: Houston, Texas, USA
- Contact:
Re: image recognition in 3.0
Hi,
maybe a better approach, than validating a whole screenshot is to validate distinctive areas of you application.
By using smaller areas you usually should be able to use higher similarity values.
Regards,
Tobias
Support Team
maybe a better approach, than validating a whole screenshot is to validate distinctive areas of you application.
By using smaller areas you usually should be able to use higher similarity values.
Regards,
Tobias
Support Team
-
- Posts: 169
- Joined: Thu Jan 13, 2011 2:06 pm
Re: image recognition in 3.0
The area I am doing image recognition is about 32x32 pixels in size. We are trying to do image recognition on a set of symbol fonts we use on the screen to represent different objects and are attempting to test them individually. I am not trying to do image recognition on an entire screen.
- Support Team
- Site Admin
- Posts: 12145
- Joined: Fri Jul 07, 2006 4:30 pm
- Location: Houston, Texas, USA
- Contact:
Re: image recognition in 3.0
Depends on the objects in the image you are trying to recognize and whether the objects are rendered differently on different systems and resolutions. E.g. if you try to validate a picture, there should not be much difference when doing so on different operating systems or resolutions.tallahassee101 wrote:Thanks that helped, though I noticed the image recognition is fairly sensitive.
If I understood you right, you try to recognize fonts? This could be more problematic, because Windows renders fonts depending on the screen resolution, DPI settings, and differently in Windows XP and Vista/7.
Regards,
Alex
Ranorex Team