The skeleton is a continuous planar shape for representation as a kind of primitive of the original. The skeleton efficiently concentrates the topological information of the original shape. It is particularly useful for representing amorphous, irregular shapes that cannot be treated by more conventional geometrical methods. The possible applications include the creation of shape primitives, curve segmentation, logging deformation history of deformable objects as well as image preprocessing for shape recognition. In this paper, we proposed a new method of skeletonization using Hough Transform and geometric constraints. We at first identify all the true and spurious skeleton branches using Hough transform and then eliminate spurious branches using two geometric constraints. The geometric constraints in our case are (i) Ratio of length between main skeleton branch & each sub-skeleton branch (ii) Angle between main skeleton branch & each sub-skeleton branch. Our experiment results are more efficient than existing works.

Digital Image Processing Skeletonization Hough Transform and Shape Matching