You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
An automatic algorithm for registration has been published in the paper "A pyramid approach to subpixel registration based on intensity" by Philippe Thévenaz, et al. The paper has however some errors which make the functions incorrect so here's the erratum http://bigwww.epfl.ch/publications/thevenaz9801.html
There's an ImageJ plugin for that (TurboReg http://bigwww.epfl.ch/thevenaz/turboreg/ ) and another that automates it for multi-frame images (Stackreg http://bigwww.epfl.ch/thevenaz/stackreg/ ). The distributed TurboReg binary for Unix seems however to have a bug and it needs to be recompiled (run Compile and run.. from the plugins menu of ImageJ. It might be needed to move the javac.jar into /usr/lib/jvm/java-6-openjdk/jre/lib/ext/ if ImageJ can't find the java compiler. The javac.jar can be downloaded from the Fiji project) . Also, it sometimes goes completely crazy giving completely abnormal results.
This algorithm seems to use the Marquardt-Lavenberg algorithm for non-linear least squares optimization which is part of the optim package of octave. They also propose a faster version of the Marquardt-Lavenberg algorithm which could maybe be added to the optim package.
The text was updated successfully, but these errors were encountered:
An automatic algorithm for registration has been published in the paper "A pyramid approach to subpixel registration based on intensity" by Philippe Thévenaz, et al. The paper has however some errors which make the functions incorrect so here's the erratum http://bigwww.epfl.ch/publications/thevenaz9801.html
There's an ImageJ plugin for that (TurboReg http://bigwww.epfl.ch/thevenaz/turboreg/ ) and another that automates it for multi-frame images (Stackreg http://bigwww.epfl.ch/thevenaz/stackreg/ ). The distributed TurboReg binary for Unix seems however to have a bug and it needs to be recompiled (run Compile and run.. from the plugins menu of ImageJ. It might be needed to move the javac.jar into /usr/lib/jvm/java-6-openjdk/jre/lib/ext/ if ImageJ can't find the java compiler. The javac.jar can be downloaded from the Fiji project) . Also, it sometimes goes completely crazy giving completely abnormal results.
This algorithm seems to use the Marquardt-Lavenberg algorithm for non-linear least squares optimization which is part of the optim package of octave. They also propose a faster version of the Marquardt-Lavenberg algorithm which could maybe be added to the optim package.
The text was updated successfully, but these errors were encountered: