Feasibility Study on Optical Image Modulation Based Parallax Generator for Monocular Visual Odometry

Abstract

We propose a method to improve the accuracy of Monocular Visual Odometry (MVO) by slowly modulating the lens using a miniature Optical Image Stabilizer (OIS) mounted on a smartphone. This method can be applied to the existing MVO or Structure from Motion (SfM) algorithm and can solve the reduction in accuracy due to the scaling issue and the lack of pixel parallax. In addition, OIS control can be performed at the same time, which can increase the accuracy against the vibration of the camera body. In this research, the lens was controlled to move horizontally with respect to the ground at a sine wave of 1 Hz without OIS control. The experimental results show that the accuracy of the monocular camera can be greatly improved.