Bioinspired Image Stabilization Control Using the Adaptive Gain Adjustment Scheme of Vestibulo-Ocular Reflex

Abstract

In this paper, an image stabilization control inspired by biological gaze stabilization mechanism is addressed. An image taken from the camera mounted on a mobile robot undergoes undesirable fluctuation that leads to poor system performance in autonomous or manual operation. Therefore, image stabilization of such camera system is essential and has been received much attention in the machine vision literature. On the other hand, a human is capable of stabilizing visual input, and this biological image stabilization mechanism is generally known as vestibulo-ocular reflex (VOR) in physiology. A particular interest of this paper is to realize this VOR features to image stabilization control problem. The mathematical interpretation of VOR in terms of the stabilization control is derived and a VOR adaptation rule is newly proposed. In the proposed method, the reference to control system is adaptively modulated by an adaptive gain to minimize residual vibration of the camera as the VOR gain converges to optimal state. To verify the proposed method, the linear shaft actuator-based image stabilization device is briefly introduced first, and it is utilized for the simulation model and experiments. Simulink simulation and experiments were conducted to demonstrate the proposed method, and results showed the proposed scheme contributes to image stabilizing effect. The image stabilization performance were validated from the vibration measurements and interframe transformation fidelity of the image sequence. From simulation and experimental results, it will be shown that the proposed method has a high potential to be applied in image stabilization control under significant system variation and uncertainties.