Linear and nonlinear generative probabilistic class models for shape contours
Proceedings of the 24th international conference on Machine learning
Measures for Benchmarking of Automatic Correspondence Algorithms
Journal of Mathematical Imaging and Vision
Probabilistic models for shapes as continuous curves
Journal of Mathematical Imaging and Vision
Shape analysis of planar objects with arbitrary topologies using conformal geometry
ECCV'10 Proceedings of the 11th European conference on Computer vision: Part V
| Hi-index | 0.00 |
Automatic construction of Shape Models from examples hasbeen the focus of intense research during the last coupleof years. These methods have proved to be useful forshape segmentation, tracking and shape understanding. Inthis paper novel theory to automate shape modelling is described.The theory is intrinsically defined for curves althoughcurves are infinite dimensional objects. The theoryis independent of parameterisation and affine transformations.We suggest a method for implementing the ideasand compare it to minimising the Description Length of themodel (MDL). It turns out that the accuracy of the two methodsis comparable. Both the MDL and our approach can getstuck at local minima. Our algorithm is less computationalexpensive and relatively good solutions are obtained after afew iterations. The MDL is, however, better suited at fine-tuningthe parameters given good initial estimates to theproblem. It is shown that a combination of the two methodsoutperforms either on its own.