RoCNet++: Triangle-based descriptor for accurate and robust point cloud registration

Karim Slimani, Catherine Achard, Brahim Tamadazte

Abstract: Highlights•Point cloud registration method based on a new descriptor encoding triangles.•Experimental validation on three well-established point clouds datasets.•Local triangles encoding enhances registration accuracy and robustness.•Efficient pose estimation using a novel farthest sampling-guided module