Image registration based on part of the image
11 次查看(过去 30 天)
显示 更早的评论
Hi all!
I want to register a pair of image based on only part of them. Is it possible in MATLAB.
One example is shown as follows. The first one is the "ground truth", the second one is unregistered image. I filled the useless part of the unregistered image using zeros so the part that needs to be registered is surrounded by black.
![](https://www.mathworks.com/matlabcentral/answers/uploaded_files/717744/image.png)
![](https://www.mathworks.com/matlabcentral/answers/uploaded_files/717749/image.png)
But when using the normal algorithm provided in registration toolbox (using command "registrationEstimator") , it fails frequently, although when cropping central part of both images, the algorithm can work smoothly. But I lost some information in that way, so that's why I fill the smaller image with black on the margins and try to register the whole image pair.
Is there any way to tell MATLAB that it only needs to focus on the center part, and apply the same transformation on the rest afterwards?
Thanks a lot!!
0 个评论
回答(1 个)
Vidip
2024-2-15
In MATLAB, you can perform image registration by specifying a region of interest (ROI) in the images, which allows the registration algorithm to focus on a particular part of the images. This can be especially useful when you have images with large regions that should be excluded from the registration process, such as the black-filled margins in your case.
By using ginput, you can focus on specific parts of the image that are important for registration, and the algorithm will ignore the black-filled margins. After estimating the transformation using these points, you can then apply it to the entire image, including the black-filled parts. This allows you to preserve the entire image content while still focusing the registration on the central, content-rich areas.
For further information, refer to the documentation link below:
0 个评论
另请参阅
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!