Abstract:

Near-infrared optical tracking systems can restore the movement of tracked objects in real time based on the reflective marker balls attached to the tracked objects. They have been widely used in many fields. This paper proposes a real-time tracking method for multiple near-infrared targets that is robust to target loss. First, according to the imaging characteristics of reflective marker balls in near-infrared cameras, the geometric center of each reflective marker ball is extracted using the grayscale centroid method. Then, the SORT algorithm is used as a multi-target tracking method in each monocular camera to match each marker point between frames. The matching relationship of the image points of the reflective marker balls in each camera is determined based on the principle of epipolar geometry combined with the weighted bipartite graph matching method, and the three-dimensional spatial coordinates of each tracked reflective marker ball are calculated in real time based on the triangulation method. Secondly, the reflective marker balls are grouped according to the spatial position relationship between the reflective marker balls during the movement process, and the reflective marker balls belonging to the same object are identified. The appearance feature vector of the tracked object and the reflective marker ball is established based on the Euclidean distance between the reflective marker balls in the same group as the matching basis for the object loss and reappearance. The tracked object that is completely lost and then reappears is rematched using the cosine distance of the appearance feature vector. Finally, the proposed algorithm is experimentally verified. The experiment shows that the tracking accuracy of the proposed algorithm can reach 0.5mm at a speed of not less than 60 frames. In addition, the lost reproduced objects and reflective marker balls can be correctly re-matched.

Key words: multi-view vision, optical tracking, stereo matching