math - Combining IMU and image tracking pose -
i improve image-based tracking using imu. imu should used in cases when i’m losing tracking (e.g. marker or image). tracking , imu working when used separately, not working anymore when used simultaneously.
imu , image based tracking providing high level pose 3x4 matrix 3x3 rotation matrix , 3x1 translation vector. pose defined in target reference frame (meaning camera position relative tracked object). imu provides 3x4 pose matrix imu is, @ moment, available rotational device tracking, therefore no translations possible after image tracking has been lost.
in case if find image trackable, store pose provided imu (lastimuposeonmarker) , pose provided image trackable (lasttrackedposeonmarker). when lose image trackable, calculate difference rotation between “lastimuposeonmarker” , current imu pose , rotate “lasttrackedposeonmarker” difference rotation.
to calculate difference rotation matrix x applying following formula:
lasttrackedposeonmarker * x = currentimupose
==> x = lasttrackedposeonmarker^-1 * currentimupose
afterwards multiply x “lasttrackedposeonmarker” new pose of camera. newpose = x * lasttrackedposeonmarker
is mathematically allowed combine 2 poses or doing else wrong?
Comments
Post a Comment