고속 평면기반 카메라추적방법을 이용한 모바일 증강현실 구현

초록

In implementing mobile augmented reality, we need a camera tracking method that satisfies two requirements, high computing speed and high robustness, simultaneously. In this paper, we propose a method that significantly improves the speed and robustness of existing plane-based camera tracking methods in four aspects. First, the proposed method computes homography by iterative refinement. Second, it exploits a speed-oriented feature tracking method. Third, it employs a fast feature matching method that is based on clustering. Fourth, it employs a method that estimates the differential camera pose between consecutive frames. Through various experiments, it was confirmed that the proposed method improves the speed and robustness of existing plane-based camera tracking methods by 37% and 10%, respectively, and we could implement augemented reality on a commercial smart phone with high fidelity at a speed of 15 Hz.

키워드