A 3D Compact Shape Descriptor based on Largest Intersection and Projection Signature
Date:
This talk presents a 3D extension of the Largest Intersection and Projection (LIP) shape descriptor. By analyzing profiles projected onto principal planes, our method extracts high-performance geometric features for 3D object classification. Validated on ModelNet and industrial datasets (tree bark defects), the approach offers a highly efficient and interpretable alternative for shape analysis with reduced feature dimensionality.
