재3차원 공간 내의 평면 객체와 비선형 칼만필터를 이용한 카메라 자세 추정 기법

다국어 초록 (Multilingual Abstract)

In this paper, we describe a method for estimating camera pose using the feature points present on a 2D planar object which is located in the world coordinate system. One of the most conventional approaches for estimating camera pose using a planar objects is to employ the 2D homography between the planar object and the image plane. In this paper, we proposes a camera pose estimation method using the Unscented Kalman Filter(UKF). The proposed method provides more robust camera pose estimation compared to the homography-only approach and minimizes issues such as error accumulation that can occur when using the homography alone. In addition, it provides improved camera tracking performance by detecting new feature points reliably even when the reference features for the camera pose initialization move completely out of the camera's field of view. The effectiveness of the proposed method is examined through several experiments and results.