3D object matching based on spherical Hilbert scanning

This paper describes a novel method to match objects in cluttered scenes. This method makes use of Hilbert scanning of feature points in Hough space. We use a 3D Hough transform to obtain a spectrum on which 3D features are concentrated on the sphere. Then, based on the obtained Hough Spectrum, we apply Hilbert scanning on the sphere to match the objects. Using this approach, we can match the object correctly and robustly in both overlapping and noise situation. The characteristic of this method is that it is a global matching method without an estimate of the rotation first and suffering from computational complexity brought by voting/correlation procedure. The experiment results show that the method is more effective compared to existing methods in both matching rate and robustness.