-
-
Notifications
You must be signed in to change notification settings - Fork 55.7k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
support similarity masks #18624
support similarity masks #18624
Conversation
Thank you for contribution! Please add simple test for new feature (reuse existed test code and testdata with adding artificial "rectangle" on input like on example). |
Added the test, I hope it's good enough |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Great! Thank you for update 👍
Could you please take a look on comments below.
If you have some issues, then I can prepare proposals in the end of week.
6f25cd7
to
d4a622c
Compare
I ended up reworking |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Great! Thank you 👍
* support similarity masks * add test for similarity threshold * short license in test * use UMat in buildSimilarityMask * fix win32 warnings * fix test indentation * fix umat/mat sync * no in-place argument for erode/dilate
Content
This PR adds the necessary code to add a similarity mask to the exposure compensator of the stitching pipeline.
You can configure a
similarity_threshold
which defines a threshold on the relative error between the pixels of two images. If the error is above the threshold, it is ignored when calculating the exposure compensation. This allows the removal of ghosting artifacts due to exposure compensation when part of the images are different (for instance a moving car which would only be present on a single picture)Note that by default the behavior is unchanged.
Example
The top 2 images are the source images. The bottom left one used a block, per channel exposure compensation. The bottom right used the same, with a similarity threshold of 0.1.
Despite the seam finder and blender doing their job perfectly, the red artifact persists due to the exposure compensator.
The red rectangle has obviously been added artificially but that kind of artifact can exist naturally (example of a moving car above).
Pull Request Readiness Checklist
See details at https://github.com/opencv/opencv/wiki/How_to_contribute#making-a-good-pull-request
Patch to opencv_extra has the same branch name.
-> The feature is not documented, but neither are the other features are the exposure compensators. They belong to the
detail
namespace, I guess this is why ?