Image-based Vision Processing for the Smart Crane Ammunition Transfer System

Image-based vision processing can be effective in detecting important geometrical features of a cargo (pallet) and lead to accurate estimation of the object s position/orientation required for grappling/pickup operations. In this paper, we investigate techniques for identifying salient features of pallet images suitable for establishing a 3D map based on stereo cameras (which would be a low-cost, low-mass alternative to a commonly used laser scanner system). Software implementation and experimental results are also presented.