Application of metaheuristic optimization algorithms for image registration in mobile robot visual control

Abstract

Visual Servoing (VS) of a mobile robot requires advanced digital image processing, and one of the techniques especially fitting for this complex task is Image Registration (IR). In general, IR involves the geometrical alignment of images, and it can be viewed as an optimization problem. Therefore, we propose Metaheuristic Optimization Algorithms (MOA) for IR in VS of a mobile robot. The comprehensive comparison study of three state-of-the-art MOA, namely the Slime Mould Algorithm (SMA), Harris Hawks Optimizer (HHO), and Whale Optimization Algorithm (WOA) is presented. The previously mentioned MOA used for IR are evaluated on 12 pairs of stereo images obtained by a mobile robot stereo vision system in a laboratory model of a manufacturing environment. The MATLsoftware package is used for the implementation of the considered optimization algorithms. Acquired experimental results show that SMA outperforms HHO and WOA, while all three algorithms perform satisfactory alignment of images captured from various mobile robot poses.