An accurate multi-sensor fusion measurement method for marker-free augmented reality system

Abstract

Augmented Reality has attracted more and more academic interests in recent years. Unfortunately there are few studies on designing a AR system for textureless target without markers. In order to break up the constraints of pre-defined markers, we propose a robust augmented reality system for textureless maker-free targets with high precision level. The performance of traditional vision-based reconstruction methods was limited when handling textureless targets. To improve this, we design a multi-sensor fusion system to reconstruct the textureless non-cooperative spherical target based on four simple lasers and a visual camera, which can provide accurate pose information for augmented reality without any depth sensor. The experiment results indicate that the our system can estimate the geometric parameters of textureless spherical target with high accuracy in real time and this method performs well on augmented reality application.