Novel 3D local feature descriptor of point clouds based on spatial voxel homogenization for feature matching

Visual Computing for Industry, Biomedicine, and Art

Table 2 Registration accuracy results of descriptors experimented on two datasets

Descriptor	Stanford repository dataset	Kinect dataset
SI	31.6	0.01
FPFH	33.4	0.02
USC	49.2	0.02
SHOT	48.3	0.11
RoPS	60.7	0.17
TOLDI	67.7	0.23
SGC	73.4	0.26
LoVS	78.2	0.28
Proposed	83.4	0.30