Object Recognition Method Using a Network Oriented Skeleton Computation

The paper describes a method (algorithm) for object recognition using skeleton computation. Skeletons represent a powerful tool for qualitative shape matching because they resume, synthesize and help the understanding both of the object shape and of its topology. On a robot system the computation of a skeleton can take many seconds, due to the relatively decreased processing power of the CPU, moreover not all robot vision systems give access to the pixel. The method describes how the processing power can be transferred from the robot controller to a high speed PC system over the network, transferring only a small quantity of information. The image is then processed, the skeletons are computed, and a matching algorithm is used to recognize the objects using the skeletons