Visual Tracking of Dynamic Overhead Cranes via Sparse Representation

Abstract Fast and accurate tracking is ultimate important to dynamic overhead cranes control. This work presents a simple yet effective fast dynamic overhead cranes tracking method, enabling it to be applied to accurate and real time controlled. Firstly, a sparse measurement matrix is utilized to acquire the features. Secondly, compressed samples are utilized in constructing sparse measurement matrix. Finally, tracking of dynamic overhead cranes is represented as a binary classification with Bayesian classifier. Experimental results on VOT2013 benchmark and self-built Cranes40 dataset show that the proposed sparse representation tracking algorithm could achieve satisfying results on tracking overhead cranes.