적외선 카메라를 이용한 헬맷 트랙커 알고리즘

Abstract

In this paper, the helmet tracker algorithm is proposed to estimate the helmets position and attitude. In particular, it operates regardless of the shape and number of obtained features. This algorithm consists of the feature detection & extraction algorithm, feature matching algorithm and 3D motion estimation algorithm. The feature detection & extraction algorithm gets the center of the feature by using the motion data from 3D motion estimation algorithm. Feature matching algorithm obtains the correspondence of the feature from the left and right image pairs. The 3D motion estimation algorithm uses the extended Kalman filter and estimates the position and attitude of the helmet by using the result of the feature detection & extraction algorithm and feature matching algorithm. The rotational experiment using the accurate rate table is conducted to verify the performance of the helmet tracker algorithm. The experimental results show that proposed helmet tracker algorithm operates even though the shape and the number of the features are varying. Thus, the helmet tracker can extend the working volume and be carried out at the dynamic movement.