Camera-based Occlusion Detection for Banknote Sorting

Abstract

Intelligent technical systems need to become more flexible and adaptive in the context of Cyber-Physical-Systems and Industry 4.0. Cash supply is very important for the worldwide economy. Within the supply chain, handling processes are highly automated, e.g., automated teller machines are installed to ensure an efficient and reliable cash supply. Recently, there is a trend in automated cash handling systems to replace the cash cassettes for each denomination by a simple pitch-in-plastic-bag for all banknotes. This necessitates new automated handling approaches because the bag content will unload irregularly. We propose a new approach which is able to detect the pose and possible occlusions of randomly distributed textured banknotes with simple smartphone cameras. This information can be used to control a handling robot-picking out banknotes.