Method for Extracting Object Skeletons

This article presents a novel method for extracting object skeletons from binary images. Unlike MATLAB's built-in bwmorph function with its 'skel' operation, our implementation utilizes a custom iterative thinning algorithm that preserves topological properties while achieving higher precision. The core algorithm works by repeatedly removing outer pixel layers while maintaining connectivity through 8-neighborhood analysis and end-point detection. Although computationally slower due to its pixel-wise processing approach, this method generates more accurate skeletal structures with better branch preservation. We will detail the algorithmic implementation using structured element templates and connectivity checks, along with MATLAB code snippets demonstrating the thinning iterations. Additionally, we explore potential performance optimizations through parallel processing and boundary tracking techniques to address the speed limitations. The implementation includes special handling for corner cases and junction points using morphological hit-or-miss transforms. This technical discussion aims to provide practitioners with a robust alternative for applications requiring precise skeletal representations.