ITK
ITK copied to clipboard
Regarding the issue of inconsistent results of the itk registration
I used the setup shown below to ensure the consistency of the sampling points. However, the result of each registration is still different.
What other operations need to be done to ensure the consistency of registration results?
Thank you for contributing an issue! 🙏
Welcome to the ITK community! 🤗👋☀️
We are glad you are here and appreciate your contribution. Please keep in mind our community participation guidelines. 📜 Also, please check existing open issues and consider discussion on the ITK Discourse. 📖
Thank you for your answer. I have now made the following changes:
(1)
(2)
The problem now found is that if the console is not turned off, the result of each registration is the same. However, if the console is turned off and the program is run again, the results will be inconsistent. Do you have any suggestions on this issue?
Thank you for your answer.I have now made some changes as above .
The problem now found is that if the console is not turned off, the result of each registration is the same. However, if the console is turned off and the program is run again, the results will be inconsistent. Do you have any suggestions on this issue?
@blowekamp @zivy do you have some suggestions?
Hello @jianjun0407, Not sure what you mean by "console is not turned off".
Based on the settings:
1. Fixed random seed.
2. REGULAR sampling strategy (note that this [includes a random location perturbation (https://github.com/InsightSoftwareConsortium/ITK/blob/a01f9fc26fd603a6adb617cc603e31ccfcd73d09/Modules/Registration/RegistrationMethodsv4/include/itkImageRegistrationMethodv4.hxx#L968-L997), but should be repeatable after the fixed random seed is set).
3. Default number of threads and mean squares metric (multiple threads shouldn't effect this metric but does effect MI).
You should get repeatable results. Possibly print the ImageRegistrationMethod to see that all the settings are as expected.