We propose a novel approach for verifying model hypotheses in cluttered and heavily occluded 3D scenes. Instead of verifying one hypothesis at a time, as done by most state-of-the-art 3D object recognition methods, we determine object and pose instances according to a global optimization stage based on a cost function which encompasses geometrical cues. Peculiar to our approach is the inherent ability to detect significantly occluded objects without increasing the amount of false positives, so that the operating point of the object recognition algorithm can nicely move toward a higher recall without sacrificing precision.

With the proposed approach, heavily occluded and cluttered scenes (left)
are handled by evaluating a high number of hypotheses (center), then retaining
only those providing a coherent interpretation of the scene according to a global
optimization framework based on geometric cues (right) (red:scene points, white: model points)

Code 

An implementation of the method is available in the Point Cloud Library, within the pcl_recognition module.

References

A. Aldoma, F. Tombari, L. Di Stefano, M. Vincze, “A Global Hypotheses Verification Method for 3D Object Recognition“, 12th European Conference on Computer Vision (ECCV), 2012 [PDF]