Local voxelized structure for 3D local shape description: A binary representation

This paper proposes a novel binary descriptor named local voxelized structure (LoVS) for 3D local shape description. Unlike many previous local shape descriptors relying on geometric attributes such as curvature and normals, LoVS simply uses point spatial locations to encode the local shape structure represented by point clouds into bit string. Specifically, LoVS is computed on a local cubic volume around the keypoint. The orientation of the cubic is determined by a local reference frame (LRF) to achieve rotation invariance. Then, the cubic is uniformly split into a set of voxels. A voxel is attached with label 1 if there are points inside, otherwise, it produces a 0 bit. All these labels therefore integrates into the LoVS descriptor. We evaluate our method on three public datasets. On each dataset, the LoVS descriptor outperforms all other descriptors tested.