Skip to Main Content
We address the problem of building detailed models of the shape and appearance of complex structures, given only a training set of representative images and some minimal manual intervention. We focus on objects with repeating structures (such as bones in the hands), which can cause normal deformable registration techniques to fall into local minima and fail. Using a sparse annotation of a single image we can construct a parts+geometry model capable of locating a small set of features on every training image. Iterative refinement leads to a model which can locate structures accurately and reliably. The resulting sparse annotations are sufficient to initialise a dense groupwise registration algorithm, which gives a detailed correspondence between all images in the set. We demonstrate the method on a large set of radiographs of the hand, achieving sub-millimeter accuracy.