doi:10.16186/j.cnki.1673-9787.2019110101

Received:2019/11/27

Revised:2020/01/13

Published:2021/03/15

Study on bobbin tracking method based on improved CamShift algorithm

XU Jian, CHEN Qianqian, HUANG Lei, SUN Zewei

College of Electronic Information，Xi'an Polytechnic University，Xi' an 710048 ， Shaanxi， China

Abstract:In order to improve the real-time and accurate tracking of bobbins， the CamShift algorithm was applied to the field of bobbin sorting and feeding tracking. The traditional MeanShift algorithm needs to manually select the initial tracking window， which makes it difficult for the tracking algorithm with a single color feature in the bobbin sorting and feeding process to meet the real-time tracking requirements. In response to the above problems， color recognition was added to the CamShift tracking algorithm to effectively reduce the mistracking phenomenon. The Kalman prediction mechanism was added to solve the problem of automatically selecting the initial tracking window. Through three sets of comparative experiments， it could be concluded that the improved algorithm kept the deviation of the target's center of mass coordinates at about 4 pixels， which verified the realtime and accuracy of the proposed method.

Key words:MeanShift algorithm;CamShift algorithm;Kalman prediction mechanism;target tracking

 基于改进CamShift算法的纱管跟踪方法研究_徐健.pdf