Output details

Planar Numerical Signature Theory Applied to Object Recognition

<23> This paper considers geometric invariants for object recognition where the objects of interest are affected by a group of transformations. Designing robust algorithms that are tolerant to noise and image occlusion remains an open problem. We demonstrate that the proposed algorithm is numerically stable and also, in terms of mean square error, a closer approximation compared to the original numerical signature formulation (Calabi). It has been applied to the classification of abnormalities in computer tomography images for detecting aneurysm. This work will be combined with a level set algorithm for the development of a new 3D image segmentation technique.