1-Attempt Subfield-Based Parallel Thinning

Abstract

A widely used skeletonization technique is thinning, which is an iterative layer-by-layer erosion in a topology preserving way. In the conventional implementation of thinning algorithms, the deletability of all border pixels in the actual picture is to be investigated. That is why we introduced the concept of 1-attempt thinning. In the case of a 1-attempt algorithm, if a border pixel is not deletable in an iteration step, it cannot be deleted in the remaining thinning phases. This paper presents a computationally efficient implementation scheme for 1-attempt 2-subfield parallel thinning; proves that a 2-subfield parallel thinning algorithm (acting on the conventional 2D square grid) is 1-attempt; shows that 1-attempt property is useful and algorithms fulfilling this property can be implemented with a remarkable speed up.