Abstract: A parallel skeleton extraction method based on a dynamic coverage strategy is proposed to address the issues of high computational complexity and low computational efficiency faced by traditional skeleton extraction methods when handling large-scale volumetric data. First, the dynamic coverage boundary propagation method effectively improves the efficiency of the skeleton extraction process. Then, the method iteratively selects candidate sphere centers on the boundary points of the inscribed spheres and obtains the final skeleton through pruning and reconstruction steps. Finally, optimization techniques such as distance transform parallel computation, reduction to find extrema, and dynamic parallel processing are introduced to further enhance computational efficiency, enabling the efficient extraction of 3D skeletons in high-resolution voxel spaces. Experimental results compared with various skeletonization methods show that the proposed method achieves a 100-fold speedup over traditional serial algorithms while maintaining high-quality skeleton extraction, making it particularly suitable for large-scale 3D data processing.