Research on Target Recognition and Tracking Technology Based on Laser Point Cloud
Long Qin, Yanxi Wei, Fan Yu, Zihao Liu · 2023
In view of the common road surface, multiple objects in a complex environment, because the detection results of multiple objects are not ideal, resulting in errors in the detection results, in order to achieve the accurate tracking of the target object. The basic principle of Camshift algorithm is emphasized, and Kalman filtering algorithm is added in the tracking process to process the obtained target data to form a moving track, predict the position of the target next time, and solve the problem of follow-up tracking. Through the verification under the complex background of the daily road, it is found that the improved algorithm can effectively and accurately extract the moving target to be tracked from the continuous frames of the point cloud image, and has high robustness. The problem of high real-time requirement for target tracking has been effectively solved. The example shows that the prediction effect is good, and it is feasible.