Approximating Binary Object Skeletonization with Pixel-Level Asynchronous Propagation

Abstract:
In this paper, we introduce a method for performing approximate skeletonization of binary images by using an asynchronous 2D binary propagation network. The application and adaptation of a previously proposed 2D single-pixel extraction network for the more complex operation of object skeletonization and the required preprocessing of the binary image are examined. Matlab simulations, with a network model which ensures correspondence with a transistor-level IC implementation, are presented. The results show that the proposed method can successfully approximate the skeletonization operation for natural image shapes and significant trade-offs in implementation complexity are possible, depending on application needs.