An Optimized L1-Medial Skeleton Extraction Algorithm

Skeleton model is the general representation form of a 3D model, which intuitively shows the topological connectivity and geometric structure of the model. As a fast and robust algorithm for extracting scattered point cloud skeleton, the L
1
algorithm has some shortcomings, such as insufficient skeleton accuracy because of the uneven distribution of local points, and poor skeleton reproducibility caused by random sampling. In this paper, we present an optimized L
1
skeleton extraction algorithm for the above problems. Firstly, on the basis of the octree-like spatial segmentation, adaptive point cloud enhancement is performed on the models with serious partial loss to balance the cloud distribution. Then, the subspace-based nearest neighbor sampling algorithm is used to downsample the enhanced point cloud to eliminate the randomness of sampling. Finally, the skeleton is extracted from the enhanced point cloud based on the resulting sampling points. The algorithm achieves efficient management of point cloud through octree-like space and improve the accuracy and timeliness of the skeleton. The adaptive point cloud enhanced sampling strategy makes the extracted skeleton more reproducible and descriptive. Experimental results show that the optimized algorithm is suitable for point cloud models in various fields and has good adaptability and robustness.