2D/3D Point landmark detection in ITK (Lloyd, Warfield)

From NAMIC Wiki
Revision as of 13:38, 18 December 2006 by Andy (talk | contribs) (Update from Wiki)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search
Home < 2D < 3D Point landmark detection in ITK (Lloyd, Warfield)

File:SlideBlloydnew.ppt

I have implemented a 2D/3D point landmark detection algorithm described by Ruiz-Alzola [1] (for scalar images only). It is based on calculating an estimate of the auto-correlation matrix at each point. From this matrix a measure of local structure can be computed. A similar approach is also used in the Harris corner detector and other landmark detection schemes.


Goal: develop the necessary ITK classes to implement the algorithm. These should be general enough to allow the creation of a family of related detection schemes to be added easily (e.g. a "blob" detector based on the Hessian).


Interface: need help in designing easy to use and clear API, and placing my class in the ITK hierarchy.


Implementation: need help accessing ITK CVS repository and checking in a filter with regression code. Also need help optimizing the implementation e.g. separable filtering, comparison of analytic derivative of Gaussian convolution and Gaussian smoothing followed by central differences.


[1] J. Ruiz-Alzola and R. Kikinis and C.-F. Westin. "Detection of point landmarks in multidimensional tensor data", Signal Processing 2001 [1]