Groupwise Medial Axis Transform

Medial Axis Transform (MAT) is one of the promising tools used for the shape recognition and it poses the advantages like space and complexity reduction, ease of processing for shape recognition. MAT representation of shapes uses object centred co-ordinate system that represents bending, elongation and thickness. But MAT of the objects is very much sensitive to small perturbations of its boundary. To overcome this problem, we prune the particular portion. In our approach, we use local or global pruning for branch significance computation. For this we use group-wise approach in which we develop a group-wise skeletonization framework that gives fuzzy significance for each branch. This is called as Groupwise Medial Axis Transform (G-MAT). This approach has several applications like shape analysis and shape recognition. This approach has been tested on various geometries of the shapes and gives good recognition results.