A Simple and Practical Solution to the Rigid Body Motion Segmentation Problem Using a RGB-D Camera

Motion segmentation with a moving camera has many applications in computer vision. This paper presents a novel method for rigid motion segmentation using a RGB-D camera. The method estimates the lengths of edges in a Delaunay graph of the feature points and observes any variation of these lengths. Non-rigid edges in the graph are identified and pruned. The resulting graph is then examined for connected components to find different motion groups. We present results which show that the approach is able to correctly identify the number of motions and their memberships on real sequences taken from a moving RGB-D camera. The method is fast and suitable for real time applications.