16/12/03: I am currently producing some AVI files that quickly reflect on the registration procedures.
18/12/03: I have investigated warping functions in greater depth. Papers still need to be read on CPS.
18/12/03: The bump will need to be modified to become a new synthetic object. This should be done within the next few days.
Background
We are investigating ways in which non-rigid registration and active appearance models are used together to surpass previous techniques.
Planned Tasks
There are several experiments that need to be run at some stage:
Real data (of anatomy) to be used rather than the simple bump.
Try to improve Ws value assignments.
Other tasks:
Speed up the existing MATLAB code.
Make the code more readable.
To Do List
This list will no longer be updated
Large Scale
Test make_1d_images.m to see how it works and generate plots of data (images) it outputs.
See how build_1d_model.m uses make_1d_images.m.
...
Reproduction of existing experiments and results.
See list in previous section and consider other alternations to the code.
...
Small Scale
Find a sensible way to automatically save images on files while the simulations run.
Find an application that is able to generate AVI clips from a collection of individual images.
This will allow the quick viewing of simulations previously run.
Get CVS access to the Division's MATLAB code (Mike Rogers).
Produce tables of registration performance.
...
Figures and Videos
I am hoping to put some visual results here to motivate better discussions about the project.